Cash Holdings of European Firms

Tilburg School of Economics and Management Department of Finance Master Thesis in Finance Cash Holdings of European Firms Author Georgi Bachurov ANR 554956 Supervisor Prof. Dr. V. P. Ioannidou July 2013

Cash Holdings of European Firms Georgi Bachurov ANR 554956 Tilburg University Tilburg School of Economics and Management Department of Finance M.Sc. Finance Supervisor Prof. Dr. V. P. Ioannidou July 2013 Abstract This paper examines the determinants of corporate cash holdings in publicly traded companies from five European countries France, Germany, Italy, Spain and the United Kingdom during 1985 and 2012. The evidence shows that the average cash ratio increases from 9.95% to 13.62% in 2012. The results suggest that cash holdings are positively affected by the market-to-book ratio, industry cash flow risk, and R&D expenses. They are also negatively related to leverage, CAPEX and net working capital. The increase in cash holdings is more pronounced in small and financially constrained firms non-dividend payers, negative net income and high cash flow volatility. This is in line with the precautionary motive for holding cash. There is mixed evidence to support the transaction and agency view. Cash holdings react to the global economic and financial crisis by experiencing higher volatility a sharp decrease followed by an increase. The European sovereign debt crisis puts additional pressure as evidenced by a steady decrease of cash holdings in the last two years of the sample period.

1. Introduction U.S. executives are alarmingly building up cash buffers instead of investing in factories and equipment 1. The accumulation of cash and cash equivalents in the first quarter of 2013 has reached a record high $1.73 trillion. At the same time, capital expenditures have risen the least since March 2010 2. There are four main motives for hoarding cash. The transaction motive suggests that firms hold cash in order to finance their operations instead of raising external capital, which is costly. The precautionary motive states that firms hold cash as a buffer to cope with unexpected adverse business shocks. The third motive is the tax one, which predicts that firms hold cash in an attempt to avoid tax payments on repatriated foreign income. The agency motive relates to the information asymmetries due to which firms prefer internal financing. Also entrenched managers accumulate more cash to pursue their own benefits. The importance of the buildup of cash holdings has become increasingly popular during the last years. In good times, a company s financial situation is not a key focus to investors. Cash flow is healthy, lending is available and access to capital markets is easy and at a low cost. However, during recessions, the picture turns quite a bit darker. When the economy is contracting, cash flow decreases, lending is difficult and access to capital markets more costly. It is during such times that a company s financial strength is the primary focus of investors considerations. Many studies have investigated the evolution of cash holdings for different periods of time and it has been widely accepted that firms gradually accumulate more cash. Bates et al. (2009) find that the average cash to assets ratio for U.S. publicly traded firms has steadily increased from 10.5% in 1980 to 23.2% in 2006. This increase is largely driven by small, non-dividend paying firms, firms underwent recent IPO listings, and firms in industries with greater increase in idiosyncratic risk. More importantly, this increase is evidenced to be related to changes in firm characteristics rather than changes in the relation between firm characteristics and cash holdings. However, these relations may have changed during the financial and European sovereign debt crises. Based on the model of Opler et al. (1999), the authors find that the decrease in inventories and CAPEX, and the increase in cash flow risk and R&D expenditures are the major contributors to the increase in the cash ratio. Other studies have also examined the trends in cash holdings for different datasets and sample periods. For example, Ferreira and Vilela (2004) analyze the determinants of cash holdings for the Eurozone countries for the period between 1987 and 2000. They find that cash is positively associated with the investment opportunity set and cash flow and negatively associated with net working capital, size and leverage. The results I present in this research are in line with prior findings. Cash levels are positively related to the market-to-book ratio and R&D expenditures, which are proxies for the investment opportunity set. 1 Cash Piles Up as U.S. CEOs Play Safe With Slow-Growth Economy, Chris Burritt, 23 May 2013, Bloomberg 2 Data is based on 2,300 U.S. companies compiled by Bloomberg 3

The increase in industry cash flow risk also relates to the increase in cash holdings. On the other side, NWC, CAPEX and leverage are negatively associated with cash holdings. Previous studies have mainly focused on U.S. companies in examining the determinants of cash holdings evolution. They have also used data for the years prior to the financial crisis. Unlike these papers, this research attempts to show whether and how European firms increase their cash levels in comparison with U.S. There are significant differences between U.S and Europe in terms of financial and legal systems. Generally speaking, European law offers companies less investor protection, ownership is more concentrated in few large shareholders and access to equity financing is more costly. These differences may lead to a deviation in European corporate cash management policies as compared to U.S. As a result, different firm characteristics may be the largest contributors to the change in cash holdings. Furthermore, this study aims to provide an insight into the evolution of cash during the recent years of economic recession. European firms were severely impacted by the burst of the global economic and financial crisis in 2008. Unlike their U.S. peers, they have also been hit hard by the European sovereign debt crisis, which continues to affect corporate earnings and worsen profitability expectations. The uncertainty about the future of the euro, combined with austerity measures taken in some European Union member states and UK, are among the reasons why European companies have been pushed back into recession in 2012. Unemployment increased, consumer spending remained weak and companies postponed investment decisions. Using a dataset of firm-year observations for all public European companies from France, Germany, Italy, Spain and UK, available on Datastream for the period 1985-2012, I test the theoretical predictions of holding cash and compare the evidence with Bates et al. (2009). I particularly choose these 5 countries because some of them (Germany, UK) are still performing well in the past 4 years, whereas others (Italy, Spain) have been impacted much harder. The results presented in this paper show that the average cash ratio has been steadily increasing for the whole sample, except for the last few years. It went up from 9.95% in 1985 and reached its peak of 16.26% in 2006. Then it dropped to 14.46% in 2008, rose again to 15.72% in 2010 and sharply decreased to 13.62% in the last two years of the sample period. Although Bates et al. (2009) document a sharper increase in cash holdings, their results are similar to mine. It has to be taken into account that my sample consists of 5 different countries. I show that the cash ratio of Italian companies, for example, steadily decreases throughout the whole sample. Spanish companies show an increase in cash holdings, but the levels remain far below average. Nevertheless, Italy and Spain represent less than 10% of all firm-year observations. Overall, I find that the accumulation of cash is mainly due to changes in firm characteristics rather than changes in the relation between cash and firm characteristics. I contribute the increase of the cash ratio to a decrease in net working capital and CAPEX, and an increase in industry cash flow risk and R&D expenditures. The volatile cash ratio during the financial crisis and the subsequent European sovereign debt crisis is in line with the precautionary motive. Firms increase their cash holdings after 2008 but the trend is inverted in 2011 and 2012 because they have consumed the cash buffer generated 4

in the previous years. The most important contributors to the change in the cash ratio during that time are the market-to-book ratio and net working capital. The remainder of this paper is organized as follows. Section 2 presents the theoretical motives for holding cash and formulates the empirical hypotheses. Section 3 describes the dataset and provides definitions and summary statistics for the main explanatory variables used in the analysis. Section 4 shows the evolution of cash holdings for the whole sample, while section 5 examines this trend for different subsamples. Sections 6 and 7 investigate the determinants of the increase in the cash ratio in details. Section 8 summarizes the findings and concludes. 2. Empirical literature review Corporate cash holdings have been thoroughly studied in the past decades. The economics and finance literature has identified 4 major motives why companies build up cash holdings. In this section I review the existing theories and empirical evidence on cash holdings determinants. 2.1 The transaction motive The transaction motive for holding cash is associated with the need of cash in order to cover operating expenditures and finance business operations (Keynes (1936)). In practice, companies could easily have a shortage of cash due to the time lag between delivering of goods and services and receiving payment or between making substantial investments and getting the subsequent return. There are several ways to cover such cash shortfalls raising funds in the capital markets, liquidating assets, reducing/postponing investments and dividends, or renegotiating payment terms to suppliers. Most of these methods incur substantial costs for raising cash. Therefore, it is often cheaper for companies to hold cash at hand. However, holding large amounts of cash comes at a price the forgone interest on cash. Opler et al. (1999) describe the optimal amount of holding liquid assets. It is given by the cross-section of the marginal cost of liquid assets curve and the marginal cost of liquid asset shortage curve. The first one is a horizontal line because these costs are assumed to be constant. The second curve is decreasing in the marginal cost of liquid asset shortage as companies need to decrease investments more or raise more outside funds. As a result, one would expect firms which incur higher transaction costs to hold more cash. Classic finance literature has shown evidence that transaction costs associated with converting noncash financial assets into cash do exists. Baumol (1952) derives the optimal demand for money by showing an increasing demand for cash when the transaction costs of liquidating assets increase. Furthermore, Miller and Orr (1966) examine higher corporate cash holdings resulting from increasing brokerage costs. Following them, Mulligan (1997) finds evidence supporting the existence of 5

economies of scale associated with the transaction motive, namely large firms hold less cash compared to smaller ones. Also, it is widely accepted that larger firms have lower probability of being in financial distress due to diversification (Rajan and Zingales 1995). Peterson and Rajan (2003) argue that fees associated with borrowing are not correlated with the size of the loan, suggesting that such costs are fixed and making it more expensive for smaller firms to raise capital externally. Therefore, I would expect a negative relation between firm size and cash levels. It is also important to note that transaction costs vary with the speed with which the company can convert noncash assets into cash. Ferreira and Vilela (2004) show that liquid assets other than cash are used as substitutes for cash because they can be easily liquidated in case of a cash shortfall. Opler et al. (1999) and Bates et al. (1999) use net working capital as a proxy for liquid assets. It reflects the firm s ability to cover its current liabilities. Bates et al. (2009) provide support for the importance of this variable as a determinant for corporate cash holdings. Since this is a cash substitute, I would expect to have a negative relation between net working capital and cash holdings. Many companies prefer to use their free cash flow as a source to undertake investments rather than cash reserves. Kim et al. (1998) build a model to predict the optimal investment in liquidity and find evidence that firms with high cash flows hold less cash because they already have high liquidity, which again serves as a cash substitute. Consequently, there is a negative relation between cash flow and cash holdings. 2.2 The precautionary motive Another extensively studied motive for holding cash, the precautionary motive, states that companies accumulate cash as a buffer against adverse business shocks. In practice, firms which hold more cash are less likely to default. Also, they can use cash for exploitation of profitable investment opportunities and reduction of financial distress costs. In the latter case, cash plays an important role as a source for investments since access to capital markets is costly. Opler et al. (1999) and later Bates et al. (2009) find evidence that cash holdings increase for firms in industries which experience an increase in idiosyncratic volatility. Campbell et al. (2001) show that there had been a significant increase in firm-level volatility. Irvine and Pontiff (2008) build on this finding and contribute the increase in idiosyncratic stock return volatility to the increase in the idiosyncratic volatility of fundamental cash flows, which is mainly due to the increased competition. Since holding cash prevents firms with more volatile cash flows from being impacted during unfavourable market conditions, there should be a positive relation between cash and cash flow volatility. A number of studies relate cash holdings to investment opportunity set as an important determinant. Ferreira and Vilela (2004) document that the level of cash holdings in European firms is positively affected by better growth options. Companies with more valuable investment projects tend to hold more cash because incurring a cash shortage is more costly for them. Opler et al. (1999) and 6

Bates et al. (2009) use market-to-book ratio and R&D expenses as proxies for investment opportunities. They argue that R&D expenses are a form of investment where information asymmetries are quite important and hence firms with higher R&D costs are assumed to have higher financial distress costs. Therefore, a positive relation is expected between those two variables and cash holdings. Capital expenditures are another variable used as a proxy for investment opportunities. Since they create assets, they could increase a firm s debt capacity and reduce the need for cash. Therefore, companies with more capital expenditures are expected to hold less cash. Another set of literature divides companies into financially constrained and unconstrained and examines their importance as determinants for corporate cash holdings. If a firm has unrestricted access to capital markets, i.e. the firm is financially unconstrained, it does not need to accumulate more cash in order to fund future investment needs. On the contrary, if firms anticipate financing constraints in the future, they tend to hoard cash today. Almeida et al. (2004) link financial constraints to a firm s propensity to save cash out of cash inflows, which they call cash flow sensitivity of cash. Their empirical results indicate that financially constrained firms have significantly positive cash-cash flow sensitivity, while unconstrained firms display no such relation. They also find support for the notion that cash holdings of financially constrained firms increase following economic recessions, while unconstrained firms do not display such tendency. Building upon these results, Han and Qiu (2007) further investigate the role of financial constraints on precautionary cash holdings. They contribute to the existing literature by showing that firms increase cash holdings when there is an increase in cash flow volatility. This behavior applies to financially constrained firms only. In contrast, financially unconstrained firms do not respond by increasing their cash levels. Now the question is how to measure whether a firm is financially constrained. There are various proxies used in previous studies. Opler et al. (1999) and Bates et al. (2009) assume that companies which do not pay dividends are riskier and have poor access to capital markets. Fama and French (2001) suggest that larger firms and more profitable firms are more likely to pay dividends. However, firms which have higher level of investment are less likely to pay dividends. In line with Bates et al. (2009), I expect a negative relation between dividends and cash holdings. Another less arguably good proxy for financial constraints is net income. Clearly, firms with negative net income are assumed to be in bad financial health, thereby increasing cash holdings in order to cope with reduced retained earnings. Lastly, leverage has been shown to influence the level of cash holdings. The sign of the relation between these two variables is not so clear to be predicted. On one hand, leverage could be positively associated with cash holdings, since firms may try to secure a cash buffer if they face higher financial distress costs (Ferreira and Vilela (2004)). Furthermore, Acharya, Almeida and Campello (2007) provide evidence for the notion that financially constrained firms prefer hoarding cash to reducing debt in the presence of valuable investment opportunities in the future. On the other hand, Ferreira and Vilela (2004) and Bates et al. (2009) document a negative relation between leverage and 7

cash holdings, suggesting that firms hold more cash to repay debt and reduce financial distress costs. Acharya, Almeida and Campello (2007) also show that financially unconstrained firms prefer reducing debt to accumulating cash. Overall, I would expect to observe a negative relation between leverage and cash holdings. 2.3 The tax motive The tax motive for holding cash becomes increasing popular nowadays. It has been reported that large US multinationals hold more cash in their international subsidiaries because they are facing huge tax expenses should they repatriate it to the home country. This results from the fact that U.S. and many other countries tax the foreign income of their firms in case of repatriation. Foley et al. (2007) provide empirical evidence that US multinationals indeed hold more cash abroad. This behavior is particularly pronounced for technology-intensive companies but does not apply to financially constrained ones. In line with Bates et al. (2009), who use firms with nonmissing foreign pretax income to measure the effect of taxes on cash levels, I would expect a positive relation between foreign income and cash holdings. It has to be taken into consideration, however, that there are significant EU taxation differences which may lead to misleading interpretation of results. 2.4 The agency motive The agency motive for holding cash stems from the conflicting interests nature between managers and shareholders. At the heart of this problem lie decisions related to payouts to shareholders. Jensen (1986) suggests that managers are inclined to accumulate cash instead of distributing it to shareholders because they increase their power and influence over the company and can extract private benefits. Also, the retained cash allows them to avoid using external financing (capital markets) and thus being able to invest in unprofitable projects without any monitoring. This typically applies to companies which generate high levels of free cash flow. One set of studies use firm characteristics as proxies for the agency costs of free cash flow which help identifying conditions that increase the probability of holding excessive amounts of cash. Opler et al. (1999) expect that companies with low market-to-book ratio and entrenched management are more likely to accumulate cash, thus hypothesizing a negative relation. Following this, Ferreira and Vilela (2004) predict that large firms have larger shareholder dispersion which in turn provides management with more freedom to hoard and spend cash ineffectively. Also large firms are less likely to be a takeover target simply because the bidder would need much more resources. As a result, a positive relation is expected between size and cash holdings. However, both of these studies do not find strong support for the abovementioned hypotheses. Bates et al. (2009) follow a different approach utilizing the GIM index as a measure of managerial entrenchment. This index combines 24 antitakeover governance mechanisms from the Investor Responsibility Research Centre. They find that 8

firms with more entrenched management hold less cash which again is inconsistent with agency view for holding cash. All these papers focus on firm characteristics in determining the role of agency costs for corporate cash holdings. There is, however, other line of research which captures the cross-country differences in legal environment, investor protection, and ownership structure and capital markets development. Previous research relates weak country-level shareholder protection with higher agency costs of managerial entrenchment (La Porta et al. (2002), Claessens et al. (2002), Klapper and Love (2004), etc.). Dittmar et al. (2003) aim to provide insight into the role of corporate governance as a determinant for cash holdings using international data. Their evidence suggests that firms in countries with poor shareholder protection hold considerably more cash than firms in countries with high levels of shareholder protection. Accordingly, Ferreira and Vilela (2004) find that firms in countries with better investor protection hold less cash. In a more recent study, Kalcheva and Lins (2007) show that when external country-level shareholder protection is weak, entrenched managers accumulate more cash. Based on previous studies (La Porta et al. (1998), Dittmar et al. (2003), Ferreira and Vilela (2004), Kalcheva and Lins (2007), Bates et al. (2009)), I deploy several proxies for investor protection. The anti-director rights index (ADRI), introduced by La Porta et al. (1998), have been criticized and revised by Spamann (2010). It takes the values between zero and five, with five representing the highest possible shareholders protection. Hence, I expect a negative relation between ADRI and cash holdings. The next variable I would test is the creditor rights index (CR), originally examined by La Porta et al. (1998). It was later revised by Brockman and Unlu (2009) and employed by Bates et al. (2009) as well. It takes the values between zero and four, with four being the highest creditor rights protection. Similarly, I would expect a negative relation between CR and cash holdings. Another proxy used by La Porta et al. (1998) and Ferreira and Vilela (2004) is the rule of law index, which estimates the law and order tradition in the country produced by the country risk rating agency International Country Risk (ICR). Its values range between zero and ten, with higher values for more tradition for law and order. Again, a negative relation is expected between the rule of law index and cash holdings. Finally, La Porta et al. (1998) document a negative relation between ownership concentration and shareholder protection. In countries with weak legal protection, large shareholders can effectively monitor managers and prevent hoarding cash. Ferreira and Vilela (2004) find strong support for the negative relation between ownership concentration and cash, which is in line with Dittmar et al. (2003) and the agency view for holding cash. 2.5 The pecking order theory The pecking order theory states that the cost of financing increases with asymmetric information. Firms prioritize their sources of financing in the following order internal funds, then 9

debt, and lastly equity issuance. Introduced by Myers (1984), it suggests that equity is a less preferred way to raise capital because when managers issue new equity, investors often believe that managers think the company is overvalued and hence play a lower value on the new equity issuance. Consistent with this view, firms retain earnings to finance new valuable investments. When retained earnings are not enough, firms stockpile cash and repay debt. It is therefore expected that firms with high cash flows will hold more cash. Leverage is negatively associated with cash because when firms have more retained earnings than what their investment needs are, they prefer to reduce debt and accumulate cash. Finally, larger firms are assumed to be more profitable and should have more cash on their balances, after controlling for investment (Opler et al. (1999)). 3. Data description This section describes the dataset and the methodology used in the empirical study. Definitions of all dependent and independent variables I use are given below as well as their summary statistics. 3.1 Sample construction I construct a panel data sample of all publicly traded European companies listed on the following markets France, Germany, Italy, Spain and the United Kingdom, from 1985 to 2012 including 3. The initial dataset comprises of all surviving and non-surviving firms that are included in Datastream for the sample period. Financial firms (SIC codes 6000-6999) are excluded from the sample because they may keep cash for other reasons than the ones investigated here, i.e. to meet capital requirements. Similarly, utilities (SIC codes 4900-4999) are also excluded because their cash holdings may be under regulatory supervision. Furthermore, firms with non-positive assets and sales for a given year are removed. Firm-year observations with missing data for any other variables are left out. The final sample consists of 49,984 firm-year observations for 5,843 unique firms. The dependent variable used in this research is the cash ratio. It has been defined in several ways throughout the economics and finance literature I reviewed in the previous section. The most common measure, however, is the ratio of cash and short-term investments divided by total assets (Bates et al. (2009)). For instance, Opler et al. (1999) use the cash-to-net assets ratio 4, where net assets equals book assets minus cash. The problem with this ratio is that it generates extreme outliers for firms whose assets are primarily comprised of cash. To fix this shortcoming, Foley et al. (2007) use the logarithm of this ratio. Finally, Bates et al. (2009) use the cash to sales ratio but document no 3 I choose 1985 as a starting date because financial data for previous years is significantly limited 4 Ferreira and Vilela (2004) also deploy the cash-to-net assets ratio 10

material change in their results. Therefore, I define the cash ratio as cash and short-term investments divided by total assets, but also reproduce some regressions using the logarithm of cash-to-net assets ratio as a consistency check. Table 1: Cash Ratio by Country This table reports summary statistics for the cash ratios across five European countries. The sample includes all Datastream non-missing firm-year observations of publicly traded firms with positive book values of total assets and sales for the period 1985-2012. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded. Cash ratio is defined as cash and short-term investments divided by total assets. N is the number of observations. Country Mean SD Min Max N France 0,1376 0,1300 0 0,9282 9819 Germany 0,1367 0,1596 0 0,9386 9564 Italy 0,1245 0,1224 0 0,9075 2988 Spain 0,0933 0,1032 0,0001 0,8236 1824 UK 0,1315 0,1582 0 0,9997 25789 Table 1 above shows the descriptive statistics for the cash ratio across the five examined European countries over the sample period 1985-2012. The mean cash ratio per country ranges from 12.45% to 13.76% with the exception of Spain. It exhibits the lowest average cash ratio of 9.33%. This is a significant difference between Spain and the rest of the countries in the sample. Ferreira and Vilela (2004) also find differences in the average cash-to-assets ratio among all Eurozone countries. They argue that such differences may arise from different accounting standards, bankruptcy laws, capital markets development, etc. They examine the period of 1987 2000 and document average cash ratio for Italy and Ireland above 20%, while it is below 10% for Portugal and Spain. 3.2 Explanatory variables definition The explanatory variables used in the empirical analysis follow Bates et al. (2009) with some exceptions. They are motivated by the transaction and precautionary motives for cash holdings. The variables used for determining the importance of the agency motive were already reviewed in detail in Section 2.4. Table A.2 in the Appendix summarizes all testable hypotheses and the expected relations between cash holdings and the explanatory variables. The first independent variable I employ is the market-to-book ratio. It is a proxy for the investment opportunity set. Strong growth opportunities are associated with more cash holdings because firms would find it costly to be financially constrained. It is calculated as book value of assets minus book value of equity plus market value of equity divided by book value of assets. The ratio of R&D expenses to sales is also a proxy for investment opportunities. Since firms with higher levels of 11

these expenditures would also have higher levels of financial distress costs, a positive relation between R&D to sales and cash holdings is expected. The ratio is set to zero when R&D expenses are missing. Another variable used in this research is firm size. Larger firms are expected to hold less cash due to economies of scale. Unlike Bates et al. (2009), I define firm size as the natural logarithm of total assets for the previous year (Foley et al. (2007)). The ratio of cash flow divided by total assets is also used in the regression analysis. Cash flow is calculated as earnings after interest, dividends and taxes but before depreciation. Higher cash flow is associated with higher cash holdings. The ratio of net working capital to total assets, where net working capital is the difference between current assets and current liabilities net of cash and short-term investments is expected to move in an opposite direction from cash holdings because it includes assets which are substitute for cash. Capital expenditures create assets which could increase debt capacity and reduce the demand for cash because of their role as collateral. This variable is defined as capital expenditures divided by total assets and is expected to be negatively associated with cash holdings. Bates et al. (2009) also employ the ratio of acquisition expenses to total assets as a substitute for capital expenditures. I do not use this variable since data on this item is not enough to be included in the statistical analysis. Leverage is measured as long-term debt plus short-term debt & current portion of long-term debt. There are two opposite theories about the relation between leverage and cash holdings. On one hand, firms with high levels of debt use cash to repay it. On the other hand, highly leveraged firms may accumulate more cash as a precautionary motive. The industry cash flow risk is seen as an important determinant for cash holdings in previous studies. It is assumed that firms with high cash flow volatility are exposed to more and stronger financial shocks. Following Bates et al. (2009), I calculate the cash flow volatility as the average of the cash flow standard deviations of firms in the same two-digit SIC codes. I also require at least four observations. Finally, two dummy variables are used. The dividend dummy takes the value of 1 when firms are distributing dividends, and zero otherwise. Positive dividend payout ratio would imply that firms are less risky with better access to capital markets. The net income dummy takes the value of one when a company reports non-negative net income, and zero otherwise. Financially constrained firms are expected to hold more cash according to the precautionary motive. In order to reduce the influence of outliers on the results of my regressions and consequent wrong data interpretation, I winsorize the extreme firm-year observations as follows. Leverage lies between zero and one. R&D to sales and capital expenditures to assets are winsorized at the 1% level. The bottom tails of net working capital to assets and cash flow to assets, as well as the top tail of the market-to-book ratio are winsorized at the 1% level. 12

Table 2 below shows the descriptive statistics for the explanatory variables I reviewed. The average market-to-book ratio is 1.5766 and the average firm has a leverage ratio of about 19%. More than the half of the companies do not have or report R&D expenses. Furthermore, 66% of them have a positive dividend payout ratio and 76% have a non-negative net income. Table 2: Explanatory Variables Summary Statistics This table reports summary statistics for the explanatory variables. The sample includes all Datastream nonmissing firm-year observations of publicly traded firms with positive book values of total assets and sales for the period 1985-2012. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded. Definitions for the explanatory variables are presented in Table A.1 (in the Appendix). N is the number of observations. Variable Mean SD 25th Pctl Median 75th Pctl N MB 1,5766 0,9696 1,0262 1,2853 1,7575 49984 Real size 11,8648 2,0939 10,4059 11,6405 13,1031 45307 Cash Flow/ assets 0,0388 0,1639 0,0262 0,0642 0,1004 49984 NWC/ assets 0,0433 0,191-0,0734 0,0347 0,1582 49984 CAPEX/ assets 0,0586 0,0644 0,0189 0,0408 0,0748 49984 Leverage 0,1934 0,1622 0,0538 0,1708 0,2941 49984 Industry sigma 0,0496 0,0237 0,0293 0,0412 0,07 49979 R&D/ sales 0,0183 0,0734 0 0 0,0021 49984 Dividend dummy 0,6647 0,4721 0 1 1 49984 Net Income dummy 0,7551 0,43 1 1 1 49984 4. Cash holdings evolution over time In this section I examine the cash ratio evolution for all the publicly traded companies across the five European countries France, Germany, Italy, Spain and the United Kingdom. I also describe the changes in the leverage and net leverage ratios for the same companies over the sample period. Table 3 shows the evolution of the described ratios from 1985 to 2012. The aggregate cash ratio, which is calculated as the sum of cash and short-term investments for all observations divided by the sum of total assets for all observations, starts from 11.16% in 1985 and decreases to 9.13% in 2012, with the lowest level of 7.41% in 2000. The next column describes the average cash ratio. It ranges from 9.95% in 1985 to 13.62%, reaching its peak of 16.26% in 2006. However, this increase is less pronounced that the one investigated in the paper of Bates et al. (2009) for U.S. companies. The steady increase between 1999 and 2006 coincides with times of global economic boost and improving business environment. It is not surprising that the average cash ratio exhibits a downward trend in the years after. The median cash ratio, although smaller in terms of size, has the same trend as the average 13

cash ratio, starting from 7.20% to 8.59% with the highest level of 10.90% in 2010. Its second highest level of 10.37% is in 2006 (see also Figure A.1 in the Appendix). In order to test if these trends are statistically significant, regressions on a constant and time are produced and the results are shown in Table A.3 in the Appendix. The coefficient on the time trend for the average cash ratio represents a yearly increase of 0.26% and has a p-value below 0.01. The R 2 of the regression model is 76.65%. This result confirms the apparent upward time trend for the average cash ratio. Similarly, the slope coefficient for the median cash ratio corresponds to a yearly increase of 0.18%, its p-value is below 0.01 and has R 2 of 64.67%. As mentioned earlier, these results indicate that the increase in cash ratio across the European companies is smaller than the one shown by Bates et al. (2009) who evidence a yearly increase of 0.46% and 0.27% for the average cash ratio and the median cash ratio respectively across U.S. companies. Columns 6 and 7 in Table 3 report the evolution of leverage ratios over the sample period. Following Bates et al. (2009), I examine the average and the median leverage ratios (see also Figure A.2 in the Appendix). The former increases from 11.51% in 1985 to 17.37% in 2012, and the latter from 9.35% to 14.35%. They both have similar trends and increase in 2008 and 2009 which might be due to the global economic and financial crisis. Next to them, I examine the average net and the median net leverage ratios, which are produced by subtracting cash and short-term investments from the total debt level. I obtain a different time trend for both of them. The average net leverage ratio exhibits an increase in the first half of the sample, starting from 1.56% to 8.78% in 1999. Then it sharply decreases, increases again in 2001 and 2002 and decreases steadily between 2003 and 2007, followed by an increase in the first and most severe years of the financial crisis. The median net leverage ratio is slightly higher but experiences almost the same trend. These results, not as strong evidence as the one shown in Bates et al. (2009), indicate that firms increase cash holdings rather than because they have reduced outstanding debt. While the average leverage ratio remains stable over the sample period, the average net leverage ratio has a downward trend. Regressions on a constant and time are also presented in Table A.3. The time coefficient for the average net leverage ratio has a decrease of 0.19% yearly, while the time coefficient for the median net leverage ratio has a decrease of 0.11% yearly. They are both statistically significant at the 1% level. R 2 for the first regression model is 33.43% and for the second 12.43%. I also examine the evolution of the average cash ratio for the different countries. Results are reported in Table A.4 in the Appendix. The average cash ratios for France, Germany and UK experience similar trends as the whole sample. They are consistently higher in the second half of the sample. The cash ratios in Germany and UK have the steepest increase and reach their peaks of 18.59% and 16.58% in 2006 respectively. Interestingly, the average cash ratio for Italy is decreasing throughout the whole sample period, starting from 16.81% in 1985 to 12.15% in 2012. The average cash ratio for Spain is slowly increasing and equals 12.97% in 2012, which is its highest level. It has to be noted, however, the observations for Italy and Spain are 4,812 altogether, whereas France, 14

Germany and UK are represented by 45,172 firm-year observations. Therefore, the evolution of cash holdings is significantly influenced by those 3 countries. Table 3: Average and Median Cash and Leverage Ratios The sample includes all Datastream non-missing firm-year observations of publicly traded firms with positive book values of total assets and sales for the period 1985-2012. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded. Cash ratio is defined as cash and short-term investments divided by total assets. Leverage ratio is defined as long-term debt plus short-term debt & current portion of long-term debt divided by total assets. N is the number of observations. Year N Aggregate Cash Ratio Average Cash Ratio Median Cash Ratio Average Leverage Median Leverage Average Net Leverage Median Net Leverage 1985 404 0,1116 0,0995 0,0720 0,1151 0,0935 0,0156 0,0144 1986 474 0,1277 0,1128 0,0867 0,1272 0,1025 0,0143 0,0143 1987 859 0,1270 0,1200 0,0800 0,1525 0,1316 0,0324 0,0390 1988 1176 0,1188 0,1113 0,0750 0,1616 0,1453 0,0503 0,0633 1989 1332 0,1151 0,1068 0,0689 0,1836 0,1713 0,0768 0,0988 1990 1416 0,1093 0,1028 0,0632 0,1997 0,1907 0,0969 0,1178 1991 1514 0,1012 0,0989 0,0568 0,2072 0,1920 0,1082 0,1216 1992 1576 0,0994 0,1018 0,0577 0,2094 0,1950 0,1076 0,1216 1993 1651 0,1037 0,1066 0,0639 0,2020 0,1838 0,0954 0,1074 1994 1748 0,1127 0,1107 0,0730 0,1888 0,1706 0,0781 0,0946 1995 1776 0,1045 0,1080 0,0659 0,1893 0,1730 0,0814 0,1011 1996 1798 0,0966 0,1085 0,0667 0,1909 0,1723 0,0824 0,0935 1997 1801 0,0931 0,1154 0,0754 0,1883 0,1728 0,0729 0,0881 1998 1885 0,0854 0,1152 0,0700 0,1960 0,1759 0,0808 0,1057 1999 1804 0,0842 0,1119 0,0645 0,1998 0,1800 0,0878 0,1053 2000 2179 0,0741 0,1392 0,0725 0,1967 0,1745 0,0575 0,0986 2001 2303 0,0772 0,1405 0,0712 0,2039 0,1817 0,0634 0,1035 2002 2439 0,0796 0,1409 0,0774 0,2097 0,1853 0,0688 0,1031 2003 2307 0,0865 0,1437 0,0834 0,2033 0,1797 0,0596 0,0954 2004 2326 0,0915 0,1524 0,0960 0,1918 0,1664 0,0394 0,0690 2005 2410 0,0926 0,1565 0,0991 0,1920 0,1628 0,0355 0,0738 2006 2539 0,0895 0,1626 0,1037 0,1889 0,1565 0,0263 0,0623 2007 2533 0,0861 0,1533 0,0972 0,1936 0,1695 0,0404 0,0682 2008 2430 0,0817 0,1446 0,0885 0,2061 0,1784 0,0615 0,0881 2009 2298 0,0973 0,1506 0,1017 0,2017 0,1788 0,0512 0,0814 2010 2209 0,1003 0,1572 0,1090 0,1896 0,1610 0,0324 0,0566 2011 2104 0,0935 0,1467 0,1000 0,1879 0,1623 0,0412 0,0588 2012 693 0,0913 0,1362 0,0859 0,1737 0,1435 0,0374 0,0661 15

5. Cash holdings evolution for certain types of firms In this section I further investigate the steady increase in cash holdings as evidenced in the previous part. To assess whether the change in cash ratios could be attributed to certain types of firms, I divide the sample using different criteria firm size, dividend and accounting performance, and cash flow volatility. 5.1 Cash ratio by firm size In order to examine whether the size of a firm could be an indicator for the cash ratio increase, I divide the sample into five quintiles by year according to the book value of total assets from the previous year. Quintile 1 represents the firms with the smallest size, whereas quintile 5 combines the largest companies over the sample. The results are presented in Figure 1 below. Consistent with Bates et al. (2009), firms in the smallest quintile experience the largest increase in average cash ratios, while firms in the largest quintile do not show any clear time trend in their average cash ratios. The increase in cash holdings is driven by smaller firms and is largely pronounced after 1999. This is the beginning of a steady increase in global economic output. Also, Fama and French (2004) show that new smaller, weaker firms and firms with more volatile cash flows become public. They are riskier in nature and access to capital markets is more costly. Therefore, it is cheaper for them to hold more cash. To verify these findings, I regress the cash ratio on a constant and time for each size quintile (results are reported in Table A.4 in the Appendix). The slope coefficients for the two smallest quintiles are positive and much higher than the ones for the last two quintiles. This confirms the upward trend in the average cash ratio for the smallest firms. The time coefficient in Model 5 is even negative, which indicates the decrease observed in the average cash ratios for the largest firms. All coefficients are statistically significant at the 1% level. 16

Cash ratio Figure 1: Average Cash Ratios by Firm Size Quintiles The sample includes all Datastream non-missing firm-year observations of publicly traded firms with positive book values of total assets and sales for the period 1985-2012. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded. Cash ratio is defined as cash and short-term investments divided by total assets. Firm-year observations are divided into quintiles based on the book value of total assets in the previous year. Quintile 1 represents the smallest firms, whereas quintile 5 represents the largest firms. 0,25 Q1: Smallest firm size quintile Q2 Q3 Q4 Q5: Largest firm size quintile 0,20 0,15 0,10 0,05 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 Year 5.2 Cash ratio by net income and dividend status As a next step, I divide the sample into firms with negative and non-negative net income. Both subsamples exhibit an increase in average cash holdings per year. Results are reported in Table 4. The cash ratio for firms with negative net income ranges from 9.89% in 1985 to 14.43%, reaching its highest level of 19.56% in 2000. Similarly, the cash ratio for firms with non-negative net income starts from 9.95% in 1985 and reaches 13.34% in 2012. However, it doesn t experience as much an increase as the cash ratio for firms with negative income. In the first half of the sample, the latter plunges to 5.71% in 1992, while in the second half of the sample it is steadily around 16-19%. It is also significantly negatively impacted in 2008 probably due to the financial crisis. As a whole, the cash ratio for firms with negative net income is much more volatile than the one for firms with nonnegative net income. This evidence is in line with Bates et al. (2009) and the precautionary motives for holding cash financially constrained firms tend to hold more cash. The regressions on a constant and time reaffirm the upward trend in the cash ratios for both firms with negative and non-negative net income. The slope coefficients are positive and statistically significant at the 1% level (see Table A.5 in the Appendix). However, the coefficient in Model 2 is twice as high as the one in Model 1, which supports the conclusion that the financially constrained firms exhibit a much larger increase. 17

Table 4: Average Cash Ratios by Accounting Performance and Dividend Status The sample includes all Datastream non-missing firm-year observations of publicly traded firms with positive book values of total assets and sales for the period 1985-2012. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded. Cash ratio is defined as cash and short-term investments divided by total assets. Firms with accounting losses are assigned to the negative net income subsample. A firm is classified as a dividend payer if it paid common dividends in the same year. Accounting Performance Dividend Status Year Negative Net Non-negative Net Nondividend Income Income Payer Dividend Payer 1985 0,0989 0,0995 0,0903 0,1002 1986 0,1226 0,1123 0,1135 0,1128 1987 0,0735 0,1229 0,1140 0,1209 1988 0,0734 0,1126 0,1212 0,1101 1989 0,0705 0,1093 0,0915 0,1084 1990 0,0625 0,1073 0,0878 0,1043 1991 0,0641 0,1069 0,0820 0,1017 1992 0,0571 0,1149 0,0732 0,1082 1993 0,0701 0,1184 0,0770 0,1148 1994 0,0789 0,1175 0,0868 0,1180 1995 0,0885 0,1119 0,0906 0,1123 1996 0,0912 0,1125 0,1039 0,1097 1997 0,1225 0,1142 0,1178 0,1147 1998 0,1171 0,1148 0,1211 0,1135 1999 0,1193 0,1101 0,1312 0,1056 2000 0,1956 0,1193 0,2122 0,1024 2001 0,1849 0,1159 0,1997 0,1036 2002 0,1637 0,1258 0,1736 0,1150 2003 0,1619 0,1334 0,1743 0,1185 2004 0,1807 0,1402 0,1848 0,1234 2005 0,1856 0,1449 0,1903 0,1262 2006 0,1902 0,1521 0,1946 0,1325 2007 0,1813 0,1435 0,1822 0,1264 2008 0,1498 0,1422 0,1702 0,1234 2009 0,1416 0,1562 0,1635 0,1383 2010 0,1581 0,1569 0,1709 0,1437 2011 0,1406 0,1490 0,1629 0,1335 2012 0,1443 0,1334 0,1547 0,1239 18

Columns 4 and 5 from Table 4 reports the evolution of the cash ratio for dividend and nondividend paying companies. A firm is classified as a dividend payer if it paid common dividends in the same year. The results are very similar in terms of direction and volatility. The cash ratio of nondividend paying firms fluctuates a lot, but it experiences a dramatic increase in the second half of the sample. It reaches its peak of 21.22% in 2000 from 13.12% in 1999. The cash ratio for dividend paying firms remains stable over time with a small increase of 10.02% in 1985 to 12.39% in 2012. The time regressions support this upward trend (see Table A.4 in the Appendix). Both Models 3 and 4 have positive and statistically significant at the 1% level slope coefficients, with the one for nondividend payers being four times as big as the one for dividend payers. Many papers support my evidence on the cash holding increases of non-dividend payers as explained by the precautionary motive. 5.3 Cash ratio by cash flow volatility Another distinctive characteristic which I examine in order to better understand the firms which have the highest levels of cash holdings is the cash flow volatility. The precautionary motive for holding cash postulates that firms in industries which exhibit the largest increase in cash flow volatility accumulate more cash. Therefore, I divide the sample into tertiles and sort the companies according to the cash flow volatility per industry and year. I use the two-digit SIC industry codes. For each firm I calculate the standard deviation of the cash flow to assets ratio for the previous three years. I require at least four observations. Then I compute the average standard deviation in each year for each two-digit SIC code industry. The evolution of the average cash ratio by cash flow volatility tertile is represented in Figure 2. As predicted, firms in the highest cash flow volatility tertile hold much more cash than firms in the other two tertiles. This increase is explicitly pronounced in the second half of the sample period. Again, the results can be largely explained by the start of a long period of economic growth and prosperity. The characteristics of the newly listed firms change and smaller, riskier and firms with more uncertain expected cash flows become public. Next to the graph, I run regressions on a constant and time to test if the upward trend in average cash ratio by cash flow volatility is statistically significant. Again, the slope coefficients are all positive and significant at the 1% level (see Table A.6 in the Appendix). Furthermore, the coefficient for the highest cash flow volatility tertile is almost five times as big as the one for the lowest cash flow volatility tertile. It corresponds to a yearly increase of 0.29% for the average cash ratio and its R 2 is 47.06%. As a result, these findings are consistent with Bates et al. (2009) and support the precautionary motive for holding cash. 19