Determinants of Corporate Cash Holdings Evidence from European Companies

Transcription

1 Determinants of Corporate Cash Holdings Evidence from European Companies A.P. Flipse* Student number: Abstract This paper investigates the determinants of cash holdings for a sample consisting of 15 European countries. The cash-to-assets ratio showed a dramatic increase from 1995 to Cash holdings slightly decreased during the recent financial crisis. Changes in firm characteristics over time are primarily responsible for the increase in cash holdings: net working capital of firms dramatically decreases, they become more R&D intensive, and cash flows become riskier. These findings provide evidence that the precautionary motive is very important in explaining European cash holdings. However, the development during the recent financial crisis cannot be completely assigned to changes in firm characteristics, since the impact of leverage, cash flow, and firm size on cash holdings significantly changes during this period. Further, this study finds that self-interested managers are, in the absence of high country-level investor protection, more likely to spend excess cash quickly rather than to hoard it. While some relevant differences exist in taxation systems between European countries, this paper does not find any proof that the tax motive is responsible for changes in cash holdings. *I am grateful to my thesis supervisor, Dr. Vasso Ioannidou, for some meaningful comments and insights on this topic. Further I would like to thank my girlfriend and parents for their support.

2 Determinants of Corporate Cash Holdings Evidence from European Companies Master Thesis of the Department of Finance Faculty of Economics and Business Administration Tilburg University Author: A.P. (René) Flipse Student number: Program: Master Finance Supervisor: Dr. V. Ioannidou Place: Tilburg Date: February 21, 2012

3 Table of Contents 1. Introduction Literature review Transaction motive Precautionary motive Tax motive Agency motive The evolution of European cash holdings and net debt Sample construction Definition of the dependent variable Cash ratio by country of incorporation Evolution of the cash and leverage ratios over the sample period The evolution of cash holdings for different subsamples Firm size and the cash ratio New issue status, dividend status, accounting performance and the cash ratio Industry cash flow risk and the cash ratio IPO cohorts and the cash ratio Cash holdings for high-tech and manufacturing firms The impact of firm characteristics on cash ratios Definition of explanatory variables Regressions estimating the determinants of cash holdings Regressions allowing for changes in intercepts and slopes over time Firm characteristics responsible for the changes in cash holdings over time Agency problems and the cash ratio Firm characteristics and the cash ratio from an agency perspective Investor protection and the cash ratio Agency problems and the value of cash The value of cash over time The value of cash for different legal origins The tax motive Conclusion References... 53

4 Appendix A Appendix B... 59

5 1. Introduction A recent article in The Guardian mentioned that data compiled by Bloomberg showed that 466 of Europe s biggest companies were sitting on 445 billion pound (691 billion dollar) of cash at the end of September It is well known that U.S. companies have also build up cash piles for many years now. For example, research of Bates, Kahle and Stulz (2009), hereafter titled as BKS, showed that U.S. firms have dramatically increased cash holdings during the period of BKS found that the average cash ratio (measured as cash and marketable securities divided by total assets) more than doubled during this period, namely from 10.5% in 1980 to 23.2% in Holding cash can be costly for firms, since investments in liquid assets (like Treasury securities) have a very low return. By investing in cash and marketable securities, a firm foregoes investment in more productive assets and therefore generates opportunity costs. Furthermore, these funds could also have been used to increase shareholders wealth by for example increasing dividends or by acquiring another firm. Trading in financial securities also generates transaction costs and negative tax consequences compared to when shareholders hold these financial securities on their own (Kim et al., 1998). Finally, agency problems seem to arise due to the high cash balances of firms. So if it can be costly for firms to hold cash, why do companies hold such a significant amount of their assets in cash, as shown by the cash ratios of BKS above? Finance and economics literature on this topic have identified four motives to hold cash. The first motive for holding cash is the transaction motive (Baumol, 1952; Miller and Orr, 1966; Mulligan, 1997; Drobetz and Grüninger, 2007; and BKS). It states that large firms tend to hold less cash, because they face lower transaction costs due to economies of scale in converting a non-cash financial asset into cash. The precautionary motive (Opler, Pinkowitz, Stulz and Williamson, 1999; Almeida, Campello and Weisbach, 2004; Ferreira and Vilela, 2004; Drobetz and Grüninger, 2007; Han and Qiu, 2007; and BKS) is the second motive and states that companies with more investment and growth opportunities hold more cash to hedge against very costly adverse shocks in cash flows. It also states that firms with riskier cash flows and poor access to the capital market hold more liquid assets. The third motive for holding cash, the tax motive (Foley, Hartzell, Titman and Twite, 2007), mentions that multinational firms hold higher levels of cash due to higher tax expenses associated with repatriation of foreign earnings. Finally, the agency motive (Jensen, 1986; Dittmar, Mahrt-Smith and Servaes, 2003; Ferreira and Vilela, 2004; Pinkowitz, Stulz and Williamson, 2006; Dittmar and Mahrt-Smith, 2007; and Harford, Mansi and Maxwell, 1 Europe s biggest companies sitting on 445bn cash, by Tom Bawden, The Guardian, 30 November

6 2008) deals with the conflicts that arise between managers and outside investors on how to deploy internal generated funds. For example, these funds could be spent internally by managers, used for acquisitions, distributed to shareholders, and/or accumulated by managers. The agency motive states that self-interested managers will trade-off the private benefits from spending these funds now against the increased flexibility of retaining these funds. Furthermore, these self-interested managers need to deal with the possible costs of being disciplined by outside investors for excess spending or accumulating too much cash. This paper will investigate the empirical predictions of these four motives for holding cash, by analyzing the cash ratios for European firms in the period of Tests will be conducted to examine whether the evolution in cash ratios over time frame can be explained by known determinants of cash holdings. Until now, only little is known about the determinants of cash holdings for European companies. Only some studies are published in this field for a sample of U.K. firms (Ozkan and Ozkan, 2004), Swiss firms (Drobetz and Grüninger, 2007) and European Economic and Monetary Union (EMU)- countries (Ferreira and Vilela, 2004). However, none of these studies investigates the tax motive for holding cash. Furthermore, these studies mainly concentrate on one or two particular motive(s) for holding cash. Investigating all four cash motives together will provide more insight in the determinants of corporate cash holdings. Finally these studies show ambiguous results concerning the relationship between agency problems and cash holdings. Finance literature could benefit from performing crosscountry research, since the previous studies mostly focused on one particular country and therefore had little variation in variables that are related to different levels of agency problems (like investor protection, legal environments and ownership structure). It is prominent that most studies focus on cash holdings of U.S. companies and such little evidence is available on this topic for European firms, since cash holdings play an important role in a firm s dividend policy, decisions regarding the capital structure of a firm and at last a firm s hedging policy against financial distress. In addition, there seem to be some vital differences between U.S. and European firms that make this topic relevant to investigate. For example, countries from Continental Europe (most European countries, e.g. France, Germany, and Spain) have different corporate governance levels compared to the Anglo-Saxon countries (e.g. U.K. and U.S.). Anglo-Saxon countries are expected to have better investor protection and therefore agency problems could be less severe in these countries. This could, following the agency motive, lead to divergence between cash balances of most European companies and cash balances of U.S. and U.K. companies. Furthermore, tax systems vary between European countries and in turn vary between European countries and the U.S., which according to the 2

7 tax motive could lead to differences in cash holdings among firms. In accordance with these assumptions, research of Dittmar et al. (2003) showed that there exists some significant cross-country variation between the liquidity levels of European companies. Median cash ratio s in 1998 (measured as cash and marketable securities divided by net assets) were for respectively The Netherlands, Spain, Germany, U.K and France; 5.0%, 5.3%, 7.3%, 8.1% and 11.1%. The evidence provided in this paper indicates that the precautionary motive is the most important determinant of European cash holdings and explains most of the changes in cash holdings over time. Also some relationships between firm characteristics and cash holdings are consistent with the transaction motive. Some evidence is supportive for the agency view, but none supports the tax motive. These results are in line with the results of BKS, since they also found that the precautionary motive was the most important determinant of cash holdings. Contrary to BKS, this study provides evidence that agency problems indeed impact cash holdings, where BKS found no proof for the agency motive. The remainder of this paper is organized as follows. Section 2 provides a review of the empirical literature on this subject. Section 3 investigates how cash and debt holdings evolved over the sample period. Section 4 considers if the evolution of cash holdings is concentrated among different subsamples or not. Section 5 examines whether the cash ratio can be determined by several firm characteristics, and whether the relationship between these characteristics and cash holdings changed over the sample period. Further, section 6 investigates if the cash ratio can be related to several indicators for investor protection and whether agency problems contributed to the change in cash holdings over the sample period. In section 7 it is analyzed if cash holding are determined by the tax motive. Finally, section 8 concludes. 2. Literature review As mentioned shortly in the previous section, finance and economics literature on this topic have identified four motives to hold cash. This section will provide a review of the empirical literature concerning these four cash holding motives. Further, it will also be analyzed how these motives are expected to influence cash holdings for European companies. 2.1 Transaction motive The first motive was already mentioned by Keynes (1936), who stated that companies hold cash because of a transaction motive. Classic models (e.g. Baumol, 1952; Miller and Orr, 1966) estimate optimal cash levels, because firms face transaction costs associated with converting a non-cash financial asset into cash. Large firms have lower transaction cost because they can benefit from economies of 3

8 scale. Therefore the transaction motive predicts that large firms hold lower levels of cash (Miller and Orr, 1966). Drobetz and Grüninger (2007) examine the cash holdings of 156 Swiss non-financial firms over the period They find a strong negative relationship between asset tangibility and the cash ratio, which indicates that firms tend to accumulate less cash if they have a lot of assets that can easily be liquidated to accumulate cash. Therefore these companies minimize the opportunity costs of liquidity. Furthermore, Drobetz and Grüninger (2007) also find supportive evidence that large firms face economies of scale in security issuances and therefore hold less cash. These results are in accordance to the research of Mulligan (1997). I expect that transaction-based requirements for cash holdings have decreased, since firms have become more efficient in handling transactions nowadays. 2.2 Precautionary motive The second motive for firms to hold cash is to hedge the firm against risks associated with possible future adverse shocks in cash flows. If access to capital markets is costly, adverse shocks in cash flows can be very costly for firms. Firms with riskier cash flows and poor access to the capital market are therefore expected to hold higher levels of cash (Opler et al., 1999). This theory also suggests that firms with better growth and investment opportunities hold more cash, since costs associated with financial distress would be higher for these firms. Multiple studies confirm this positive relationship between the investment opportunity set, often measured as market-to-book ratio and/or R&D-spending, and corporate cash holdings (Opler et al., 1999; Ferreira and Vilela, 2004; Ozkan and Ozkan, 2004; Harford et al., 2008; and BKS). Almeida et al. (2004) found that unconstrained firms do not invest in cash out of cash flow, while on the contrary constrained firms do invest in cash out of cash flow. Han and Qiu (2007) extend this model of Almeida et al. (2004) in which they analyze the relationship between a firm s financial constraints, cash flow volatility and cash holdings. This theoretical model discusses that a financially constrained firm, in contrast to unconstrained firms, increases its cash holdings in response to increased cash flow volatility. Han and Qiu (2007) tested this relationship for a large set of U.S. publicly traded firms between 1997 and 2002 and found supportive evidence that the impact of cash flow volatility on cash holdings depends on a firm s financial constraints. Furthermore, Acharya, Almeida and Campello (2007) develop a model which mentions that in case of a low correlation between investment opportunities and operating performance, firms retain cash instead of using cash flows to reduce debt. By transferring funds from high cash flow states to low cash flow states, firms are also able to fund investment opportunities in low cash flow states. Literature shows multiple studies that investigated the determinants of corporate cash holdings which confirmed the existence of this so called precautionary motive. Opler et al. (1999) find supportive evidence for the precautionary motive using a sample of 4

9 public U.S. firms in the period BKS found that this motive for cash holdings played the most important role in explaining the increase in cash ratio for public U.S. firms in the period There is also some supportive evidence for the precautionary motive from research using a European sample. Drobetz and Grüninger (2007) find some evidence for Swiss companies, since they found that firms tend to hold more cash when the probability of financial distress increases. Ferreira and Vilela (2004) showed that firms in EMU-countries with better developed capital markets tend to hold less cash. It is difficult to predict if the precautionary demand for cash will be higher or lower during this sample period. On the one hand, developments like the expansion of the derivatives market, improvements in forecasting and control and the usage of credit lines are likely to decrease the precautionary demand for cash. On the other hand, research of Kearney and Potì (2008) showed an increase in idiosyncratic risk for European countries that are a member of the European Union over the period The same development was shown for the U.S. (Campbell, Lettau, Malkiel, and Xu, 2001). Irvine and Pontiff (2007) found that an increase in idiosyncratic risk positively impacts cash flow volatility. The increase in idiosyncratic risk could therefore imply an increase in unhedgeable risk and in turn increase the precautionary demand for cash. Especially newly listed firms are expected to have higher levels of idiosyncratic risk (Brown and Kapadia, 2007) and therefore these firms are also predicted to hold higher levels of cash. 2.3 Tax motive An additional motive for firms to hold cash is the tax motive. Companies that have foreign earnings could face negative tax consequences accompanied with the repatriation of these earnings (Foley et al., 2007). Repatriating earnings from affiliates operating in countries with a lower tax rate means relatively higher tax expenses and as a consequence these affiliates hold higher levels of cash than other affiliates. Therefore this theory predicts that multinational firms hold higher cash balances. Most empirical research on this topic has ignored the possible influence of this tax motive. BKS however included the tax motive in their research, but did not find evidence that cash holdings increased more during the sample period for firms with foreign pretax income. As mentioned shortly in the introduction of this thesis, there are some important differences between Europe and the U.S. that could result in divergence between the results of BKS and this study. European countries and the U.S. have dissimilar tax systems that could influence the level of cash holdings. De Mooij and Ederveen (2003) mention that the U.S. and particular European countries (U.K., Spain, Greece and Ireland) adopt a tax credit system. A credit system (also called as worldwide taxation) credits tax liabilities of the host country against the tax rate of the home country of the parent company. 5

10 On the contrary, the remaining European countries that are member of the European Union adopt an exemption system (or territorial taxation). This tax system exempts foreign income of the subsidiary for taxation in the home country of the parent company if these earnings are already taxed in the host country. As mentioned before, the tax motive expects multinational firms to hold higher cash balances due to the (often) negative tax consequences associated with the repatriation of foreign earnings of their subsidiaries. However, most European countries do not have this tax incentive and are therefore expected to hold lower cash balances compared to countries that have adopted the tax credit system. 2.4 Agency motive The fourth and final motive is the agency motive and deals with problems associated with the separation of ownership and control of a company. The relationship between agency problems and cash holdings is not as clear as with the previous cash holding motives. As mentioned in the introduction, conflicts could arise between managers and shareholders concerning the distribution of internal capital. Cash holdings could, for example, be spend internally, distributed to shareholders, used for corporate takeovers or used to pile up cash holdings. The free cash flow theory (see e.g. Jensen, 1986) states that, in the absence of good investment opportunities, entrenched managers are more likely to build up cash balances rather than to increase the payouts to shareholders. Supportive empirical evidence for this theory comes from cross-country research by Dittmar et al. (2003) and Ferreira and Vilela (2004). They showed that firms operating in countries with smaller agency problems, indicated by superior investor protection, hold less cash. In countries with a weaker corporate governance structure investors are less protected and therefore agency problems are expected to be more severe. However, the study of Harford et al. (2008) found that firms hold less cash in countries with a weaker corporate governance system. As an explanation for these contradicting results Harford et al. (2008) states that self-interested managers will trade-off the private benefits from spending excess cash now against the increased flexibility of retaining these funds. Moreover, these self-interested managers need to deal with the possible costs of being disciplined by outside investors for excess spending or accumulating too much cash. Large excess cash balances increases the attention of outside investors and the potential for managers to be disciplined by other stockholders. Supportive evidence for this theory is provided by Faleye (2004), since the amount of proxy fights increases in the amount of excess cash holdings, and as a result of these events the payouts of shareholders and executive turnover increases. Therefore, the threat for disciplining could cause managers to become more cautious in piling up cash balances. As a consequence, weakly controlled managers spend more funds on capital expenditures and acquisitions in order to hold lower cash balances. The profitability of these investments can often only be 6

11 valued many years after the investment and therefore limits the chance of being disciplined on the short term. This view is also in accordance to the results of Mikkelson and Partch (2003), since they demonstrated that firms with higher cash balances (firms are matched by size and industry), which following the free cash flow theory indicates more agency problems and lower firm performance, perform similarly or even better than firms holding normal levels of cash. Furthermore, multiple studies (e.g. Pinkowitz et al., 2006; Dittmar and Mahrt-Smith, 2007; and Kalvecha and Lins, 2007) showed that greater agency problems between managers and outside investors negatively impacts firm value via a lower valuation of cash holdings. Finally, conflicts could arise between minority and large shareholders. In the absence of good investor protection, large shareholders are expected to make decisions that enable them to gain higher private benefits from control. Since it is easier for these shareholders to extract cash from the company than fixed assets, these companies are predicted to hold higher cash balances (Pinkowitz et al., 2006). However, literature also demonstrates that many studies do not find evidence for the agency motive as determinant of cash holdings. For example, Opler et al. (1999) and BKS did not found evidence for the agency motive for their sample consisting of large U.S. firms. Additional empirical research showed similar results for datasets consisting of small U.S. firms (Faulkender, 2002) and U.K. firms (Ozkan and Ozkan, 2004). An explanation for these results could be that these studies mostly focus on one particular country and therefore have little variation in variables that are related to different levels of agency problems (like investor protection, legal environments and ownership structure). By using international data it could be easier to find empirical evidence for the existence of an agency motive for holding cash (Ferreira and Vilela, 2004). Besides differences in taxation systems between European countries and the U.S., variation in corporate governance levels between Anglo-Saxon countries and Continental Europe could lead to variation in agency problems. The U.S. and the U.K have a market-based system, which is characterized by dispersed ownership of companies, case law emphasizing enforcement of shareholder rights, welldeveloped stock markets and managers are restricted in this system via the market of corporate control and regulation (Goergen, Martynova and Renneboog, 2005). However, countries from Continental Europe have a stakeholder-oriented system. This corporate governance system is characterized by concentrated ownership and control, codified law focusing on the protection of stakeholders, illiquid stock markets and finally this system tries to resolve the misalignment between managers and shareholders via the monitoring by stakeholders of the firm (large shareholders, creditors or employees). Some argue that the market-based system is superior to the stakeholder-based system (La Porta et al., 7

12 1997; Levine, 1998, 1999), since it estimated that this system gives better investor protection, cheaper and easier access to financing and enhances long-term growth. As mentioned above, corporate governance systems that provide less investor protection are expected to create more severe agency problems. Therefore companies operating in Continental Europe are therefore, all else equal, expected to value cash less compared to firms incorporated in The U.S. and The U.K. In accordance to the results of Pinkowitz et al. (2006), the value of cash should also decrease over our sample period if agency problems cause the cash holdings to increase. However, Martynova and Renneboog (2008) also point out that not all governance systems are homogenous for all European countries, but instead depend on their legal origin. Martynova and Renneboog (2008) demonstrated that countries of English legal origin continuously outperformed the other countries on both the shareholder right index and minority shareholder index during the period of However, Martynova and Renneboog (2010) notice a convergence of corporate governance strategies and rules towards a marked-based system for countries of French and German legal origin. Scandinavian countries however, still demonstrate the most properties of a stakeholder-based system. These Scandinavian countries have especially lagged to reform corporate control with regulations that would enhance (minority) shareholder protection. Again, theory predicts that especially during the 1990 s agency problems are more severe for countries of Scandinavian, German and French legal origin. However towards the end of my sample period, agency problems are expected to decrease for German and French countries and therefore these countries should, ceteris paribus should value cash more than countries of Scandinavian legal origin. 3. The evolution of European cash holdings and net debt This section shows how the sample is constructed that is used for this study and how the dependent variable is defined for this research. Further, it will provide descriptive statistics on the cash ratio for each sample country and statistics on how cash holdings and leverage ratios have developed over the sample period for the whole sample. 3.1 Sample construction The data sample for this research is constructed from the WRDS Compustat Global database for the period This sample period makes it possible to analyze the developments of the cash ratio and leverage ratio during the recent financial crisis. The data includes both surviving and non-surviving is chosen as first year, since one variable for this research (cash flow risk) based on data of the prior six years and Compustat Global started with tracking European firms from

13 firms that appear in the database during this period. Exchange rates, market values and base dates 3 are obtained from the DataStream database. Exchange rates were necessary to convert data into a single currency, namely United Kingdom pound 4. Furthermore, the 3-month Euribor rates are gathered from DataStream as a proxy for the risk-free rate of European countries. Inflation rates, which are based on the consumer price index (CPI), are obtained from the World Bank database. Firms that have missing- or negative values for total assets and/or net sales are excluded from the sample. In accordance to the study of BKS, financial firms and utilities are excluded from the sample. Financial firms (SIC codes ) could hold cash for reasons which are not considered here, like meeting capital requirements. The cash holdings of utilities (SIC codes ) in turn can be subject to regulatory supervision and could therefore negatively influence results. The previous section showed that agency problems are expected to differ between European countries depending on their legal origin. To test whether firms hold cash due to an agency motive, the sample is therefore constructed of European countries that belong to the English, French, German or Scandinavian legal origin. Hereby, the categorization of La Porta et al. (1997) will be followed. The United Kingdom and Ireland belong to the English originated countries. Sample countries with a German origin are Germany, Austria and Switzerland. Furthermore firms from Spain, France, The Netherlands, Belgium, Portugal and Greece belong to the French legal origin. Finally, four sample countries belong have a Scandinavian origin, namely Sweden, Norway, Denmark and Finland. All the above requirements yielded a sample of firm-year observations for unique firms. 3.2 Definition of the dependent variable This study will investigate the determinants of cash holdings by analyzing the cash ratio. Literature demonstrates several ways to measure the cash ratio, namely cash to assets, cash to net assets and the logarithm of cash to net assets. The fist ratio is measured as cash and marketable securities divided by total assets. This measure is commonly used in literature (see e.g. BKS and Ozkan and Ozkan, 2004). The cash to net assets is used by Opler et al. (1999) and defines net assets as total assets minus cash and marketable securities. However, this variable creates extreme outliers for firms with a high proportion of assets in cash (BKS). This was indeed the case for the sample used for this research. Finally, the last ratio is defined as the logarithm of the previous ratio and is used by Foley et al. (2007). This variable reduces the problem of extreme outliers, but does not eliminate it completely. Therefore this research mainly 3 Base date is a DataStream variable that represents the date when a company enters the database. This is commonly used as an indicator for IPO-dates (see e.g. Kohl and Schaefers, 2010), since firms often enter the database after they went public. This variable is used here as a proxy for a firm s listing date, since Compustat Global generated a lot of missing values regarding listing dates for European companies. 4 The U.K. pound exchange rate was the only exchange rate that covered the complete sample period and all different currencies. 9

14 focuses on the cash ratio as defined by BKS (cash to assets). The logarithm of cash to net assets is also included in some regressions of section five and six, but then it is explicitly mentioned. 3.3 Cash ratio by country of incorporation Table 1 summarizes descriptive statistics about the cash ratio for each country in which the sample firms are incorporated. The overall mean cash ratio of the sample is 15.3%. Column 1 illustrates that companies with a French legal origin all have a lower average cash ratio than the total sample mean. In particular, Portugal, Greece and Spain (all three have French legal origin) have relatively low mean cash ratio with respectively 5.9%, 8.7% and 9.5%. The two highest mean cash ratios are both from Table 1: Descriptive statistics of cash ratio by country of incorporation The sample includes all Compustat Global firm-year observations for companies that are incorporated in Austria, Belgium, Switzerland, Germany, Denmark, Spain, Finland, France, United Kingdom, Greece, Ireland, Netherlands, Norway, Portugal or Sweden for the period In this table the countries are categorized by their legal origin following La Porta et al. (1997). Firms are excluded from the sample if they have missing or negative values for total assets and/or net sales. Financial firms (SIC codes ) and utilities (SIC codes ) are also excluded from the sample. These requirements yielded a sample of observations for unique firms. The cash ratio is measured as cash and marketable securities divided by the book value of total assets. Column (6) shows the number of firm-year observations over the sample period. Country English legal origin Mean (1) 25th perc. (2) Median (3) 75th perc. (4) Standard deviation (5) United Kingdom Ireland German legal origin Germany Austria Switzerland French legal origin France Netherlands Spain Belgium Greece Portugal Scandinavian legal origin Sweden Norway Denmark Finland N (6) 10

15 Scandinavian origin, namely Sweden (18.0%) and Norway (17.6%). Also both countries with an English origin hold relatively more cash (16.3% and 17.1%). The German legal origin countries have values close to the overall mean cash ratio. Median cash ratios of column 3 provide similar results. High standard deviations of column 5 also indicate that there is a lot of variation in the cash ratios. The mean cash ratios of all subgroups are statistically different from each other at the 5%-level, with the exemption of the mean cash ratio of the English and German legal origin. The mean cash ratios of these particular groups are only statistically different from each other at the 10% level. 3.4 Evolution of the cash and leverage ratios over the sample period Columns 2 and 3 of Table 2 demonstrate how the average and median cash ratio evolves for the whole sample over the sample period. The average cash ratio increases from 11.6% in 1995 to 15.6% in 2010, peaking in the year 2006 with 17.4%. The average cash ratio in 2010 is 134.5% of the average cash ratio in However, the cash ratio does not increase continuously and can be divided roughly into three stages. First the cash ratio increases from 1995 to 2000, then it drops with 1.5% and increases again until 2006, where after it first decreases, but then remains relatively constant at the sample s average for the remaining part of the sample period. However, average cash holdings are in the last stage still, on average, higher than during the period Although the median cash ratios in column 3 are considerably lower than mean cash ratios, they demonstrate a similar trend. The cash ratios are significantly lower for this European sample compared to the U.S. sample of BKS. For their sample, the average cash ratio increased from 17.1% to 23.3% between 1995 and Although cash ratios are considerably lower for the European sample, cash holdings increased at a higher degree during this period than for the U.S. sample (50% vs. 36%). Table 2: Average and median cash and leverage levels from The sample includes all Compustat Global firm-year observations for companies that are incorporated in Austria, Belgium, Switzerland, Germany, Denmark, Spain, Finland, France, United Kingdom, Greece, Ireland, Netherlands, Norway, Portugal or Sweden for the period Firms are excluded from the sample if they have missing or negative values for total assets and/or net sales. These requirements yielded a sample of observations for unique firms. Cash ratio is measured as cash and marketable securities divided by the book value of total assets. Leverage is measured as long-term debt plus debt in current liabilities, divided by the book value of assets. Net leverage is computed as the difference between total debt and cash and marketable securities, divided by the book value of total assets. Column (1) shows the minimum number of firm-year observations available to measure the variables of column (2) to (7). 11

16 Year N (1) Mean Cash Ratio (2) Median Cash Ratio (3) Mean Leverage (4) Median Leverage (5) A regression (not reported) is executed to test if this trend in cash holdings is also statistically significant. Hereby the cash ratio is regressed on a constant and time measured in years. The regression results provide supportive evidence for a significant increase in cash holdings, since the coefficient corresponds to a yearly increase in average cash ratio of 0.24%. The coefficient has a p-value of and the R 2 of the regression is equal to 51%. The regression results for the median cash ratio show a slightly lower slope coefficient of 0.16% with a p-value lower than The R 2 of this regression was equal to 61%. These regression results are however only useful to demonstrate the evolution in cash holdings for this sample period and not to extrapolate this trend to future years. Mean Net Leverage (6) Median Net Leverage (7) Columns 4 to 7 of Table 2 show how leverage ratios have developed over the sample period. The leverage ratios of column 4 and 5 are measured as the ratio of debt to the book value of total assets, where debt includes long-term debt and debt in current liabilities. The mean leverage ratio slightly increases from 21.9% in 1995 to 23.0% in 2010, peaking in 2003 with 26.9%. However the leverage ratio does not seem to show a distinctive time trend, since it goes upwards and downwards and therefore does not seems to be related to the trend in cash holdings. The median leverage ratio slightly decreases over the sample period from 19.4% to 17.5%. Firms seem to hold more leverage during the recent financial crisis. An estimated regression for the leverage ratio provides more evidence that leverage does 12

17 not seem to show a distinctive trend, since the time coefficient for both mean and median leverage ratios was not significant. However if cash is treated as negative debt, it could be more meaningful to analyze firms leverage by using net leverage. Net leverage is measured as total debt minus cash and marketable securities, divided by the book value of total assets. The results for average and median net leverage ratios are reported in columns 6 and 7. Although the leverage ratio increased during the sample period, the average net leverage ratio decreased with 26.2% from 10.3 to 7.6%. Therefore, if cash can really be treated as negative debt, this gives more insight in how leveraged firms are. However, average net leverage does not show such a distinctive time trend as for the U.S. sample of BKS. This is confirmed by the high p-value (0.330) of the estimated time coefficient. The median net leverage ratio also decreases from 11.0% to 8.3%. 4. The evolution of cash holdings for different subsamples The previous section demonstrated an increase in cash holdings and a less distinctive downward trend in net leverage for European firms during the sample period. This section will now analyze how widespread this development is by analyzing cash ratios among different subsamples of firms. 4.1 Firm size and the cash ratio At first, this section will analyze whether firm size influences the average cash ratio. The sample firms are for every year divided into quintiles based on the book value of their assets, which are first translated into U.K. pounds, in the prior fiscal year. Figure 1 shows how the average cash ratio develops for each size quintile over the sample period. The first quintile (Q1) represents the smallest firms in the sample and the fifth quintile (Q5) represents the largest firms in the sample. In the starting year all size quintiles hold about the same amount of cash. However, after 1995 the average cash ratio increases significantly for the first three size quintiles. For example, the cash ratio of the smallest firms (Q1) increases over the sample period from 12.6% to 22.2%. The average cash ratio of the second and third quintiles increases slightly less than the first quintile. In contrast, the average cash ratio for large firms remains constant (Q4) or even decreases (Q5) over the sample period. The largest firms hold significantly less cash than the smallest firms in 2010, since they have a cash ratio of 11.4% compared to 22.2%. The start of the recent financial crisis also has a substantial impact on the cash ratios of all size quintiles, since the average cash ratio decreases across all size quintiles over the period Regressions on a constant and time resulted in a positive time coefficient for the first four quintiles (respectively 0.45%, 0.45%, 0.27% and 0.09%) and a negative coefficient for the fifth (-0.07%) size quintile. However, the time 13

18 coefficient of the fifth size quintile was not statistically. The four remaining coefficients all had a p-value below 0.02 and are therefore statistically different from zero. These results provide evidence that large firms are not responsible for the increase in the average cash ratio during the sample period. On the contrary, the average cash ratio decreases for relatively large firms and the increase in cash holdings is only evident for small firms. Furthermore, the results show that the level of cash holdings decreases with firm size. These results are consistent with literature on the precautionary motive. Figure 1: Average cash ratio by firm size quintile from 1995 to 2010 The sample includes all Compustat Global firm-year observations for companies that are incorporated in Austria, Belgium, Switzerland, Germany, Denmark, Spain, Finland, France, United Kingdom, Greece, Ireland, Netherlands, Norway, Portugal or Sweden for the period Firms are excluded from the sample if they have missing or negative values for total assets and/or net sales. Financial firms (SIC codes ) and utilities (SIC codes ) are also excluded from the sample. These requirements yielded a sample of observations for unique firms. Cash ratio is measured as cash and marketable securities divided by the book value of total assets. The sample firms are divided into five quintiles according to their firm size that particular fiscal year. Hereby firm size is measured as the logarithm of the book value of assets (in U.K pounds). The first quintile (Q1) represents the smallest firms in the sample and the fifth quintile (Q5) represents the largest firms in the sample. Cashratio Year Q1 Q3 Q5 Q2 Q4 4.2 New issue status, dividend status, accounting performance and the cash ratio Secondly, it is considered whether the increase in cash ratio is due to the considerable increase in IPO activity during the 1990s. Firms that just went public could hold more cash because of the IPO and because these firms are likely to issue seasoned equity within several years of the IPO (BKS). To investigate whether this is the case for the European sample, firms are classified as IPO firms in a particular year if they have gone public within the prior five years and to the non-ipo subsample otherwise. Columns 1 and 2 of Table 3 demonstrate how the average cash ratio develops for both 14

19 subsamples over the sample period. The average cash ratio increases with 35.0% from 14.6% to 19.7% for firms that had just gone public during this period. For non-ipo firms the average cash ratio also increased during the sample period, namely from 11.2% in 1995 to 14.6% in However, therefore the increase (30.3%) in cash holdings is slightly lower for firms that did not go public within the 5 prior years. The average cash ratio peaks for both subsamples in the year 2005 with 24.2% for IPO firms and only 15.3% for non-ipo firms. In this year the average cash ratio was therefore 65.7% higher than the cash ratio of 1995 for newly listed firms. Although not reported here, median cash ratios show similar trends. When the time trend is estimated by regressing both cash ratios on a constant and time, both Table 3: Average cash ratios from 1995 to 2010, delineated by new issue status, dividend status and accounting performance The sample includes all Compustat Global firm-year observations for companies that are incorporated in Austria, Belgium, Switzerland, Germany, Denmark, Spain, Finland, France, United Kingdom, Greece, Ireland, Netherlands, Norway, Portugal or Sweden for the period Firms are excluded from the sample if they have missing or negative values for total assets and/or net sales. Financial firms (SIC codes ) and utilities (SIC codes ) are also excluded from the sample. These requirements yielded a sample of observations for unique firms. Cash ratio is measured as cash and marketable securities divided by the book value of total assets. Firms are classified as IPO firms in a particular year if they have gone public within the prior five years and to the non-ipo subsample otherwise. Hereby, the DataStream variable BDATE (base date) is used as a proxy of IPO-date. Firms are assigned to the subsample dividend payer if they paid common/ordinary dividend in that particular year. At last, firms are classified as negative EBIT if they have reported a negative EBIT (earnings before interest and taxes) at the particular fiscal year end. Differences in average cash ratio between the subsamples are all statistically different from zero at better than the 1% level, except of the differences in dividend status and issue status for the year 2001 and in dividend status for the year New Issues Dividend Status Accounting Performance Year IPO Firms Non-IPO Firms Dividend Payer Non-Dividend Payer Negative EBIT Non-Negative EBIT (1) (2) (3) (4) (5) (6)

20 show a significant positive time trend. However, the time coefficient for IPO firms is 0.35% (p-value 0.018) and for non-ipo firms slightly lower with 0.23% (p-value of 0.000). Therefore the documented increase in cash holdings can be partly assigned to capital raising activities of IPO firms. These results differ from the ones by BKS, because for their U.S. sample the cash ratio increased more for non-ipo firms than for IPO firms over their sample period. Therefore BKS did not find evidence that capital raising activities of IPO firms are responsible for the increase in cash holdings. Column 3 and 4 of Table 3 display how the average cash ratio develops over the sample period for firms that paid common/ordinary dividend in a particular year (labeled as dividend payer) and firms that did not (labeled as non-dividend payer) 5. The free cash flow theory of Jensen (1986) implies that agency problems cause non-dividend paying firms to accumulate more cash when these firms have poor growth opportunities. Furthermore, it is expected that dividend paying firms are more closely monitored and therefore face lower transaction costs and also have better access to the capital markets (Almeida et al., 2004; and Drobetz and Grüninger, 2007). Ferreira and Vilela (2004) add to this context that dividend paying firms could also lower dividend payouts to raise funds. As a consequence, these firms are expected to have lower precautionary requirements for cash holdings with respect to non-dividend paying firms (Han and Qiu, 2007). The results in column 3 and 4 illustrate a dramatic increase in cash holdings for the non-dividend paying subsample over the sample period. Mean cash ratios more than doubled (9.0% to 18.1%) over the sample period for the non-dividend paying firms. However, this increase is more pronounced for the years until 2000, since it then peaked with 21.9%. After 2000, the cash ratio remains relatively constant. In comparison, the average (median) cash ratio of dividend paying firms only increased with 17.7% (19.9%) during the same period from 11.8% to 13.9%. These results provide evidence that non-dividend paying firms were partly responsible for the increase in cash holdings during the period Since non-dividend paying firms are expected to be more financially constrained, these results indicate that the increase in cash holdings is more pronounced for financially constrained firms and supports the precautionary motive. However, the increase in cash holdings from 2001 to 2006 cannot be explained by the precautionary motive, since the cash ratio not significantly increases more for the non-dividend paying subsample than for the dividend paying sample. Columns 5 and 6 of Table 3 show how a firm s accounting performance impacts cash ratio. Firms that have reported negative EBIT (earnings before interest and taxes) are more likely to be financially 5 After 2006 Compustat Global did not seem to assign zeros to firms that did not pay common/ordinary dividends, since after that year all firms were labeled as dividend payers. Therefore if a firm-year observation was missing for the variable common/ordinary dividends, it is replaced with zero if another Compustat Global variable (Cash Dividends) was equal to zero. 16