Global and Regional Corporate Cash Holding Trends:
Before and After the Financial Crisis
Master’s Thesis
Author: Dian Grancharov
Student Number: 1939459
Supervisor: Drs. Ing. Nanne Brunia
University of Groningen
26.06.2014
Abstract: This master’s thesis investigates whether the increase of cash holdings over time and the differences between firms in the US, UK, and EMU regions, can be explained and predicted by fundamentals. The results confirm previous empirical research and provide evidence that the cash holding behaviour of firms can be explained by the trade-off concept of corporate structure. Exceptions arise after the crisis. EMU firms with high research and development expenses hold less cash after 2008. Moreover, after the crisis UK firms that pay out dividends hold more cash than non-dividend paying firms. The motivation for these changes is that the precautionary motive for cash is excessively strong after the crisis. Cash holdings decrease in 2008, but increase abnormally from 2009 until 2012. This thesis provides evidence for dynamic effects in cash holdings and is the first to provide strong empirical statistically significant results for the positive relationship between capital expenditures and cash.
Keywords: Cash holdings, Firm characteristics, Firm size, Precautionary theory, Trade-off theory, Pecking order theory, Capital expenditures, Financial Crisis
1
1. INTRODUCTION
During the past couple of decades, considerable attention has been devoted to the increasing levels of cash holdings held by firms worldwide. According to a recent article1 in the Wall Street Journal, the non-financial companies in the S&P500 hold $650 billion of cash overseas. According to an article in The Telegraph called “Cash Reserves Rise to £166 billion as Businesses Play it Safe”2
, businesses built up their cash reserves as defensive buffers with the onset of the recession and the financial crisis, rather than investing in organic growth, carrying out takeovers and mergers, or distributing the free cash flow to investors. The aggregate cash of the sample firms in this thesis has nearly doubled between 1991 and 2012. Appendix A reports, that in 1991 24% of all sample firms could repay all of their debt with the cash at hand. In 2012, 39% of the sample firms had more cash than debt.
Most of the previous research focuses on the empirical determinants of corporate cash holdings and mainly on firm-specific characteristics. Research has differentiated between firms in the US (Bates et al., 2009; Opler et al., 1999), firms based in any country that is a part of the European Monetary Union (Ferreira and Vilela, 2004), and UK firms (Ozkan and Ozkan, 2004). This thesis is the first to analyse systematically the differences between these three regions in terms of cash holding levels and determinants. While nearly half of the UK firms in the sample of this thesis could repay all their debt with cash in 2012, only a quarter of the EMU firms had similarly large cash holdings. Moreover, only Pinkowitz et al. (2013) have provided empirical evidence on the abnormal increase of cash holdings after the financial crisis of 2008. The research question of this master’s thesis is whether the increase of cash holdings over time and the differences between firms in the US, UK, and EMU regions, can be explained and predicted by fundamentals.
Two theoretical models help to explain which firm characteristics influence cash holdings decisions. On the one hand the static-trade off theory extends the transaction and precautionary motives of Keynes (1936) and Baumol (1952), stating that firms set target debt-to-value ratios and gradually move towards it (Myers, 1984). The trade-off model postulates that firms identify their optimal level of cash holdings by weighing the marginal costs and marginal benefits of holding cash (Ferreira and Vilela, 2004). Holding more cash reduces the likelihood of financial distress, allows the pursuance of investment policy when financial constraints are met, and minimises the costs of raising external funds or liquidating existing
1
“Breaking the Buck on Corporate Cash Piles”, Justin Lahart, 12 May 2014, The Wall Street Journal
2
assets, while also incurring an opportunity cost of the capital invested in liquid assets (Ferreira and Vilela, 2004). On the other hand is the pecking order framework, according to which, firms prefer internal to external financing, and debt to equity if issuing securities, and have no well-defined target debt-to-value ratios (Myers, 1984). The pecking-theory states that, if the internally-generated cash flow is less, then the firm draws first from its cash balance or marketable securities portfolio (Myers, 1984). If external financing is required, then firms prefer to issue debt rather than equity.
This thesis tests the empirical determinants of cash with simple panel data OLS models. Dummy variables control for the differences between regions and periods. Forecasting models based on the pre-crisis period allow for calculating the cash forecasting in the years after the crisis. Separate regional-specific models are calculated for optimal forecasting power.
The results of this thesis are especially relevant for decision making about the capital structure of firms with high or low excess cash at hand. This thesis can help managers when they need to motivate their decisions on cash holdings, and answer why excess capital is not reinvested in organic growth, takeovers or mergers, or why the free cash flow is not distributed to investors.
This thesis continues with an analysis of the theoretical concepts which explain why firms would hold up such high levels of cash. Section 3 documents the data and the descriptive statistics, and explains the methodology of this thesis. Section 4 provides and discusses the results of the statistical tests. The last section of the paper concludes.
2. LITERATURE REVIEW
3
exchanges. Keynes (1936) further classifies it as a combination of the income-motive and the business-motive. Firms hold hash with an income-motive to link the interval between the receipt of income and its payment. On the other hand, according to Keynes (1936), firms hold cash because of a business-motive, in order to bridge the interval between the time of incurring business costs and that of the receipt of the sale proceeds. Another motive for liquidity preference is the precautionary-motive, which Keynes (1936) defines as the desire for security as to the future cash equivalent of a certain proportion of total resources. The precautionary-motive for holding cash is mostly evident in emergencies requiring sudden expenditures or in the event of unforeseen opportunities of advantageous purchases.
Baumol (1952) extends the transaction-motive of Keynes (1936) by introducing a “broker’s fee”. This fee covers all non-interest costs of borrowing or making a cash withdrawal, including opportunity losses which result from having to dispose of assets just at the moment the cash is needed, losses involved in the poor resale price, administrative costs, and psychic costs (the trouble involved in making a withdrawal). According to Baumol (1952), a rational individual will, given the price level, demand cash in proportion to the square root of the value of his transactions and choose the method for meeting payments in such a way, as to minimize costs.
Table I
Relationship of Firm-specific Characteristics and Cash Holdings According to Theory The first column reports the firm-specific characteristics that are investigated in this thesis. The second and third columns report the expected relationship between the firm-specific characteristics and cash holdings according to the trade-off theory and pecking order theory, respectively. A positive expected relationship is marked with a ‘+’. A negative expected relationship is marked with a ‘-‘.
Firm-specific characteristic Trade-off theory Pecking order Theory Industry volatility + Market-to-book + + Size - +
Operating Cash Flow - +
Net Working Capital -
Capital Expenditures + +
Leverage +/- -
Research and Development + +
Dividend Dummy -
4
the firm-specific fundamentals which are covered in this thesis and cash holdings, according to the two theories.
On the one hand, the static-tradeoff theory extends the transaction and precautionary motives of Keynes (1936) and Baumol (1952), stating that firms set target debt-to-value ratios and gradually move towards it (Myers, 1984). On the other hand, the pecking order framework proposes that firms prefer internal to external financing, and debt to equity if issuing securities, and have no well-defined target debt-to-value ratios (Myers, 1984).
According to Robichek and Myers (1966), following the static-trade off concept, the amount of optimum leverage is at the point where the present value of the tax rebate associated with a marginal increase in leverage is equal to the present value of the marginal cost of the disadvantages of leverage. For example, the firm may incur operating inefficiencies and may be forced to forego investments which would have been feasible and profitable if the firm had avoided bankruptcy by borrowing less (Robichek and Myers, 1966). Moreover, Robichek and Myers (1966) prove that leverage is disadvantageous if present borrowing requires additional financing in some future contingencies and the future cost of this financing is uncertain. Scott (1976) provides empirical evidence on the optimal level of debt being an increasing function of the liquidation value of the firm’s assets, the corporate tax rate, and the size of the firm. According to Robichek and Myers (1966) the value of a firm is a concave function of the debt-to-equity ratio. Modigliani and Miller (1961) argue that a firm should be as levered as possible, if it were to decrease its taxes to a minimum. There is an opportunity to increase wealth by tax-minimizing devices, since tax liability can be reduced by a high debt/equity ratio (Hirschleifer, 1966). However, infinite debt-equity ratio is inconsistent with both common sense and established practice (Scott, 1976). As argued by Hirschleifer (1966), an optimal balance of debt and equity financing would require an integration of the personal-tax and corporate-tax effects, and consideration of other factors, such as the magnitude of bankruptcy penalties.
5
According to Baumol (1952), the demand for cash rises less than proportional with the volume of transactions, which infers economies of scale in the use of cash. Moreover, Titman and Wessels (1988) argue that larger firms are more diversified and thus less likely to experience financial distress. Therefore, all else equal, larger firms are expected to hold less cash. Mulligan (1997) empirically proves that larger firms hold less cash, for a sample of 12,000 US firms between 1961 and 1992. Titman and Twite (2007), Kim et al. (1998), Opler et al. (1999), Pinkowitz and Williamson (2001), Ferreira and Vilela (2004), and Bates et al. (2009), present more empirical evidence that cash decreases with firm size.
Kim et al. (1998) provides empirical evidence for a sample of US firms between 1975 and 1994, that firms with better investment opportunities and more volatile cash flows hold more cash. The static trade-off theory predicts that firms with better investment opportunities have greater financial distress costs because the positive NPV of these investments disappears in case of bankruptcy (Ferreira and Vilela, 2004). Opler et al. (1999) find that, for a sample of US firms between 1971 and 1994, smaller firms, with more volatile cash flows, poor access to external financing, and better investment opportunities hold more cash. Titman and Twite (2007) present similar results for a sample of US firms from 1982 to 2004. Bates et al. (2009) confirms the same results for a sample of US firms between 1980 and 2006. Hanlon (2014) confirms the same results for a sample of US firms between 1995 and 2011. Consistent with the precautionary demand for cash, the available literature on cash holdings provides abundant empirical evidence that firms with more volatile operating cash flows hold more cash.
Opler et al. (1999) and Bates et al. (2009) provide statistically insignificant results on the relationship between capital expenditures and cash for US firms. Capital expenditures are used, similarly to market-to-book ratio, as a proxy for future investments. Although, being classified as expenditures, managers expect a positive return on the capital invested. According to the precautionary and trade-off theory, firms with positive future investment opportunities hold more cash as a security precaution.
6
A firm paying dividends can raise funds at a minimum cost, by just reducing its dividend payments. All else equal, firms that pay dividends hold less cash than firms that do not pay dividends (Opler et al., 1999; Hanlon, 2014; Kim et al., 1998; Ferreira and Vilela, 2004; Bates et al., 2009).
Liquid assets, other than cash, can be used as a substitute for cash, and liquidated in the event of a cash shortage (Ferreira and Vilela, 2004). Ceteris paribus, firms with more liquid asset substitutes are expected to hold less cash (Opler et al., 1999; Bates et al. 2009; Ferreira and Vilela, 2004; Ozkan and Ozkan, 2004). Operating cash flows are another low-cost source for liquidity. Firms with higher operating cash flows are expected to accumulate less cash.
The pecking order theory of Myers (1984) has a different view on the capital structure of firms. In the well-functioning markets of Modigliani and Miller (1958), securities can always be sold at a fair price. This means that the net present value of selling securities is always zero, since the cash raised exactly balances the present value of the liability created (Myers and Majluf, 1984). Hence, managers should accept every positive NPV project, regardless of the type of financing, external or internal. However, in a study of a sample of large corporations, Donaldson (1961) reports reluctance for equity issuance, even in the case of high price-to-earnings ratios. According to Myers and Majluf (1984) as long as managers invest in every project they know to have positive NPV, nothing fundamental would change in a firm’s indifference between internal and external financing. However, if managers have inside information and act in the interests of current shareholders, there are cases in which no shares are going to be issued, even if positive NPV decisions are neglected. This happens when the cost to old shareholders of issuing shares at a bargain price outweighs the project’s NPV (Myers and Majluf, 1984). Myers and Majluf (1984) find that not issuing shares and, therefore not investing in positive NPV projects, means that management has misallocated real capital investment and ultimately firm value is reduced.
7
securities portfolio (Myers, 1984). If external financing is required, then firms prefer to issue debt rather than equity. According to Myers (1984), firms start with debt, then possibly hybrid securities such as convertible bonds, then equity as a last resort.
The pecking order theory states that the leverage ratio of each firm reflects its cumulative requirements for external finance (Myers and Majluf, 1984). Hence, changes in debt ratios are not driven by an attempt to reach an optimal capital structure (Shyam-Sunder and Myers, 1999). Goyal (2003) criticises the pecking order theory, by presenting empirical evidence of net equity issuance on average commonly exceeding net debt issuance in a sample of publicly traded American firms during the 1980s and 1990s. Moreover, Goyal (2003) is able to prove that net equity issues track the financing deficit much more closely than do net debt issues.
According to the pecking order theory, a large opportunity set of positive net present value investment would create a positive demand for cash. Therefore, consistent with the static trade-off theory, the pecking order theory predicts that firms with better investment opportunities will hold more cash.
In a pecking order world debt is issued only when all retained earnings and cash holdings combined are depleted or not enough to finance the whole investment opportunity set. This inverse relationship infers that firms with high leverage should be low on cash. According to Opler et al. (1999), larger firms are presumably more successful and have higher operating cash flows. In a pecking-order more internally generated cash means higher cash holdings. Therefore bigger firms and firms with high operating cash flows are expected to hold more cash.
8
3. DATA AND METHODOLOGY
The sample of this thesis consists of all active and dead listed firms from 1991 to 2012, based in the United States, the United Kingdom, and any country participating in Stage Three of the EMU as of 1st of March, 2014. All data is downloaded from Worldscope Datastream. Of the 28 EU Member States on the 1st of March 2014, 17 (Belgium, Germany, Estonia, Ireland, Greece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovenia, Slovakia and Finland) have adopted the euro, meaning that they participate fully in Stage Three of the EMU. The sample requires that firms have a positive invested capital and positive book value of equity, in order for those firms to be included in any given year. Financial firms with SIC codes between 6000 and 6999 are excluded because they may carry cash to meet statutory capital requirements rather than for the economic reasons studied here. Utilities, SIC Codes 4900-4999 are also excluded, because their cash holdings can be subject to different regulatory supervision in different regions. Table II provides the raw data names and symbols, Datastream mnemonics, and description of each type of financial data that is downloaded from Datastream and used in this thesis. Each firm-year observation should have data for all required datatypes, excluded otherwise. Appendix B presents the number of firm-year observations per year for the sample and each region separately.
Table II
Datatype Names, Symbols, Mnemonics, Types, and Description
Description of the data downloaded for the construction of the dependent and independent variables for this thesis. The first column reports the name of the datatype, as it is referred to in this thesis. The second column reports the symbol of the data type, as it is referred to in this thesis. The third column reports the DataStream Mnemonic of each datatype, required for the data download from Datastream. The last two columns provide the type of the data and description, according to the Worldscope Database Datatype Definition Guide.All datatypes are downloaded for the period between 1991 and 2012 for firms based in the US, UK, or any country fully participating in stage three of the European Monetary Union as of 1st of March 2014, which includes Belgium, Germany, Estonia, Ireland, Greece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovenia, Slovakia and Finland. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded from the sample. Only firms with positive invested capital and book value of equity are included. Firm-year observations that miss an observation for any datatype are excluded.
Name Symbol
Datastream
Mnemonic Type Description Book Value
of Assets
BVA WC 02999 Asset Data TOTAL ASSETS represent the sum of total current assets, long term
9
investments, net property plant and equipment and other assets.
Book Value of Equity
BE WC 03501 Shareholders'
Equity Data
COMMON EQUITY represents common shareholders' investment in a
company.
Market Value of Equity
MVE WC 08001 Stock Data MARKET CAPITALISATION Market Price-Year End * Common
Shares Outstanding Earnings Before Interest, Taxes and Depreciation
EBITDA WC 18198 Income Data EARNINGS BEFORE INTEREST, TAXES AND DEPRECIATION (EBITDA) represent the earnings of a
company before interest expense, income taxes and depreciation.
Total Interest Expense
TIE WC 01075 Expense Data TOTAL INTEREST EXPENSE includes but is not restricted to: Interest
expense on debt (WC 01251) and Interest capitalized (WC 01255)
Common Dividends
D WC 05376 Cash Flow
Data
COMMON DIVIDENDS (CASH) represent the total cash common dividends paid on the company's common stock during the fiscal year, including extra and special dividends.
Book Value of Cash and Equivalents
BVC WC 02005 Asset Data CASH & EQUIVALENTS represents Cash & Due from Banks for Banks,
Cash for Insurance companies and Cash & Short Term Investments for all
other industries.
Net Working Capital
NWC WC 03151 Supplementary Data
WORKING CAPITAL represents the difference between current assets and
current liabilities.
Capital Expenditures
CAPEX WC 04601 Cash Flow Data
CAPITAL EXPENDITURES represent the funds used to acquire fixed assets
other than those associated with acquisitions.
Book Value of Debt
BVD WC 03255 Liability Data TOTAL DEBT represents all interest bearing and capitalized lease
obligations.
Research and Development
Expense
RD WC 01201 Expense Data RESEARCH AND DEVELOPMENT EXPENSE represents all direct and indirect costs related to the creation and development of new processes, techniques, applications and products
with commercial possibilities.
Income Taxes T WC 01451 Expense Data INCOME TAXES represent all income taxes levied on the income of a company by federal, state and foreign
10
Table III documents the construction of the dataset. It includes the name and symbol of each variable, as well as a short description, and the calculation equation. Most of the variables are calculated consistently with previous literature (Opler et al., 1999; Bates et al., 2009) with several notable exceptions. Cash is typically scaled by total assets or sales in the literature. In this study cash is scaled by invested capital, because comparing cash ratios based on total assets in time and across companies is hampered by changes and differences in operating liabilities. This problem is easily solved by scaling cash by invested capital. Furthermore, finance separates operating and financing activities in order to facilitate the analysis. Therefore, all variables are scaled by invested capital excluding (excess) cash. Separating operating and financing activities does not mean that financing does not matter.
Table III
Variable Names, Symbols, Description, and Equations
The first column includes the names of the variables as they are referred to in the main body of this thesis. The second column reports the symbol of each variable as referred to in the tables of this thesis. The third column provides a description of the construction of each variable. The final column provides the equation by which each variable is calculated, using the datatype symbols reported in Table I. The variable equation for industry volatility is the typical standard deviation estimation. For the market-to-book equation calculation market value of debt is assumed to be equal to book value of debt and market value of cash is assumed to be equal to book value of cash. For the calculation of SIZE, invested capital is deflated to 2012 Euro using the Consumer Price Index. Size is used in log form to account for the wide dispersion of invested capital over firms. CASH, MB, CF, NWC, CAPEX and LEV are winsorized at the 1% and the 99% level to control for outliers.
Name Symbol Description Equation
Cash CASH Book value of cash and securities to invested capital net of cash and
equivalents.
Industry Volatility
VOL The cross-sectional standard deviation of operating cash flows of
all firms j, in industry i, for year t.
Market-to-Book
MB Market-to-Book ratio of invested capital net of cash and equivalents.
Size SIZE Natural log of invested capital net of cash and equivalents in 2012 Euros.
Cash Flow CF Operating cash flow to invested capital net of cash and equivalents
Net Working Capital
NWC Net working capital net of book value of cash and equivalents to invested
11
capital net of cash and equivalents.
Capital Expenditures
CAPEX Capital Expenditures to invested capital net of cash and equivalents.
Leverage LEV Leverage as book value debt to invested capital net of cash and
equivalents.
Research and Development
RD Research and development to invested capital net of cash and
equivalents.
Dividend Dummy
DUMDIV Dividend Dummy equal to 1, in the year in which a company pays out
dividends, zero otherwise.
The market-to-book ratio is calculated by the ratio of the market value of invested capital excluding cash, to the book value of invested capital excluding (excess) cash. This negates the effect of operating liabilities on the proxy for profitable future investments, as well. For the market-to-book equation calculation market value of debt is assumed to be equal to book value of debt and market value of cash is assumed to be equal to book value of cash. Furthermore, the size of firms is calculated by taking the natural logarithm of invested capital net of (excess) cash. Size is used in log form to account for the wide dispersion of invested capital over firms.
12
Table IV
Average and Median Cash from 1991 to 2012
The table reports average and median cash (as calculated in Table II) for each sample year for the whole sample of firms in the second and third columns. The fourth and fifth columns report the average and median cash for firms based in the US. The fifth and sixth columns report the average and median cash for firms based in the UK. The last two columns report the average and median cash for firms based in any country fully participating in stage three of the European Monetary Union as of 1st of March 2014, which includes Belgium, Germany, Estonia, Ireland, Greece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovenia, Slovakia and Finland. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded from the sample. Only firms with positive invested capital and book value of equity are included. Firm-year observations that miss an observation for any datatype are excluded.
Whole Sample US
UK EMU
Year average median average median average median average median 1991 0.248 0.075 0.282 0.094 0.243 0.054 0.199 0.070 1992 0.387 0.084 0.631 0.124 0.200 0.068 0.171 0.066 1993 0.293 0.085 0.397 0.116 0.223 0.077 0.184 0.070 1994 0.352 0.096 0.455 0.113 0.246 0.099 0.204 0.074 1995 0.276 0.084 0.324 0.091 0.251 0.096 0.176 0.072 1996 0.360 0.090 0.467 0.106 0.219 0.083 0.155 0.074 1997 0.387 0.092 0.498 0.108 0.236 0.085 0.155 0.075 1998 0.395 0.085 0.482 0.091 0.277 0.086 0.163 0.070 1999 0.316 0.072 0.374 0.073 0.225 0.070 0.177 0.071 2000 0.472 0.086 0.598 0.096 0.333 0.081 0.220 0.075 2001 0.516 0.094 0.655 0.118 0.511 0.082 0.185 0.080 2002 0.525 0.106 0.708 0.142 0.379 0.095 0.188 0.076 2003 0.441 0.102 0.545 0.164 0.424 0.110 0.208 0.084 2004 0.472 0.145 0.589 0.187 0.425 0.140 0.222 0.090 2005 0.494 0.144 0.623 0.194 0.474 0.140 0.221 0.093 2006 0.465 0.133 0.581 0.170 0.481 0.156 0.203 0.074 2007 0.464 0.126 0.605 0.164 0.434 0.147 0.185 0.077 2008 0.472 0.121 0.638 0.169 0.381 0.129 0.185 0.068 2009 0.450 0.141 0.605 0.212 0.378 0.131 0.184 0.077 2010 0.436 0.154 0.569 0.223 0.393 0.145 0.192 0.077 2011 0.434 0.144 0.576 0.208 0.369 0.138 0.190 0.078 2012 0.430 0.142 0.588 0.194 0.298 0.123 0.220 0.099
13
However, Appendix A shows that in 1991 17% of the firms in the EMU region could repay all their debt only with their cash at hand, while in 2012 one in every four firms had negative net leverage, calculated as leverage net of cash. The percentage of firms that have enough cash to repay all their debt has more than doubled for UK firms during our sample period, reaching 48% in 2012.
The results of a division of the firms in deciles and their respective median cash holdings are reported in Table VI. Each variable in each year is divided in deciles from lowest (10%) to highest (100%). The median of the respective cash holdings is calculated. Then the median for each decile across all sample years is calculated. The same method is performed for the whole sample and separate regions. The resulting relationships between the variables and cash holdings, summarized in table V, are similar between regions. It is clear that firms with high market-to-book ratios hold more cash, which is consistent with both the trade-off theory and the pecking order theory. Firms in the lowest 10% size decile hold the most cash, which is consistent with the trade-off theory, but contradicts the pecking order theory. Firms with higher capital expenditures hold more cash, which according to both theories is correct.
Table V
Relationship between Firm-specific Characteristics and Cash Holdings According to Theory and Univariate Statistics
The first column reports the firm-specific characteristics that are investigated in this thesis. The second and third columns report the expected relationship between the firm-specific characteristics and cash holdings according to the trade-off theory and pecking order theory, respectively. The last column reports the results from the univariate statistics of this thesis. A positive expected relationship is marked with a ‘+’. A negative expected relationship is marked with a ‘-‘. Firm Characteristics Trade-off theory Pecking order Theory Univariate Statistics Industry volatility + + Market-to-book + + + Size - + - Cash Flow - + +
Net Working Capital - -
Capital Expenditures + + +
Leverage +/- - -
Research and Development + + +
Dividend Dummy - -
Table VI
Division of the Variables in Deciles and the Respective Median Cash
Each variable in each year is divided in deciles from lowest (10%) to highest (100%). The median of the respective cash holdings is calculated. Then the median for each decile across all sample years is calculated. The same method is performed for the whole sample and separate regions. The sample includes all Datastream firm-year observations from 1991 to 2012 with positive invested capital, based in the US, UK or any of the 18 countries participating in the EMU as of March 2014. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded from the sample. Panel A reports the deciles based on the whole sample and the US. Panel B reports the deciles based on the UK and EMU.
Panel A
Whole Sample
Decile SIGMA MB SIZE CF NWC CAPEX LEV 10% 0.065 0.085 0.424 0.347 0.361 0.062 0.584 20% 0.076 0.052 0.214 0.112 0.112 0.078 0.293 30% 0.088 0.056 0.169 0.093 0.078 0.086 0.167 40% 0.081 0.057 0.134 0.062 0.070 0.086 0.099 50% 0.110 0.068 0.111 0.054 0.084 0.094 0.075 60% 0.128 0.076 0.097 0.069 0.090 0.104 0.058 70% 0.219 0.103 0.090 0.080 0.092 0.113 0.054 80% 0.095 0.160 0.079 0.121 0.105 0.122 0.058 90% 0.146 0.299 0.068 0.215 0.130 0.165 0.063 100% 0.177 0.796 0.055 0.678 0.178 0.282 0.162 USA
Decile SIGMA MB Size CF NWC CAPEX LEV 10% 0.074 0.106 0.485 0.727 0.531 0.074 0.715 20% 0.071 0.052 0.314 0.219 0.194 0.090 0.438 30% 0.070 0.055 0.228 0.083 0.094 0.095 0.289 40% 0.101 0.062 0.215 0.064 0.082 0.109 0.132 50% 0.165 0.073 0.175 0.065 0.099 0.107 0.087 60% 0.241 0.090 0.131 0.078 0.120 0.140 0.068 70% 0.244 0.137 0.109 0.105 0.123 0.174 0.053 80% 0.148 0.236 0.079 0.163 0.139 0.182 0.047 90% 0.424 0.462 0.067 0.306 0.157 0.234 0.063 100% 0.174 0.984 0.060 0.864 0.250 0.325 0.188
15
process of the writing of this thesis, whether or not the relationship between the independent and dependent variables is linear, is not tested for. It can be considered also as a limitation of previous research and a topic for future research. This thesis assumes that the relationship between the independent and all dependent variables is linear.
Panel B
UK
Decile SIGMA MB Size CF NWC CAPEX LEV 10% 0.076 0.066 0.490 0.330 0.366 0.076 0.316 20% 0.074 0.044 0.211 0.104 0.116 0.079 0.197 30% 0.090 0.049 0.179 0.070 0.077 0.090 0.130 40% 0.089 0.053 0.117 0.064 0.069 0.090 0.103 50% 0.114 0.074 0.101 0.063 0.081 0.110 0.071 60% 0.126 0.098 0.086 0.068 0.092 0.112 0.061 70% 0.114 0.131 0.101 0.085 0.096 0.120 0.064 80% 0.132 0.171 0.090 0.120 0.098 0.122 0.069 90% 0.131 0.245 0.088 0.189 0.129 0.152 0.085 100% 0.109 0.619 0.061 0.562 0.144 0.292 0.203 EMU
Decile SIGMA MB Size CF NWC CAPEX LEV 10% 0.081 0.061 0.151 0.096 0.116 0.044 0.157 20% 0.051 0.049 0.115 0.051 0.070 0.055 0.107 30% 0.052 0.054 0.097 0.053 0.057 0.071 0.089 40% 0.066 0.060 0.080 0.052 0.055 0.072 0.071 50% 0.083 0.064 0.071 0.062 0.066 0.071 0.064 60% 0.096 0.074 0.064 0.066 0.064 0.071 0.060 70% 0.073 0.078 0.062 0.075 0.067 0.081 0.059 80% 0.070 0.091 0.071 0.089 0.077 0.080 0.062 90% 0.109 0.118 0.058 0.123 0.095 0.098 0.065 100% 0.108 0.225 0.065 0.309 0.149 0.151 0.125
Appendix B Table 1 provides results on the yearly comparison of firms paying research and development expenses, and firms that have research and development expenses zero. There is a clear positive trend in each region, with firms paying research and development keeping more cash. Appendix B Table 2 provides results on the yearly comparison of dividend and non-dividend paying firms. Dividend paying firms consistently keep less cash, which proposes a negative relationship between dividends and cash.
16
data OLS model, which has cash as a dependent variable, and all other variables from table II as independent variables, including the dummy for dividend and non-dividend paying firms. The second model re-estimates the first model, but adds the lagged cash ratio as an independent variable. According to Brooks (2009), lagged values of the dependent variable captures important dynamic structure in the dependent variable and is likely to reduce, and possibly remove serial correlation issues, if present. Significant dynamic effects might be caused by inertia, since many variables in economics and finance change only slowly (Brooks, 2009).
To test for differences before and after the crisis, the sample is divided in three sub-sample periods. This thesis assumes the start of the crisis to be at the time of the Lehmann Brothers default. This renders the post-crisis period to be between 2008 and 2012. Pinkowitz et al. (2013) define cash holdings in the 1990s as normal and states that abnormal cash holdings of multinational firms increase sharply in the early 2000s. However, Pinkowitz et al (2013) do no prove empirically the exact year of increase. Bates et al. (2009) divides his sample years in a period that ends in 1999 and a period between 2000 and 2006, without mentioning any reasoning for his choice. This thesis, allows the dataset to divide itself in the pre-crisis period. This is performed in model 3, which is a panel data OLS model with firm and year fixed effects. According to the fixed effects in model 3, the abnormal cash hoarding started in 2001, therefore the base period is defined as the period until 2000, the pre-crisis period is between 2001 and 2007, and the post-crisis period is between 2008 and 2012. The third model in the results section includes dummy variables for the different regions to test for differences between regions. Moreover, dummy variables for the pre- and post- crisis period test for differences before and after the crisis. Interactions between the dummy variables and the independent variables allow for not only changes in the intercepts but also the slopes of the independent variables. Since this thesis is only interested in differences between the pre- and post- crisis periods, the test that allows for interactions between dummy variables and independent variables is performed only for the years between 2001 and 2012.
17
This thesis tests whether there is abnormal cash in each year of the post-crisis period compared to the pre-crisis period via a forecasting model estimated over the pre-crisis period, by running Fama-MacBeth cross-sectional regressions and taking the averages of the coefficients. Then the difference between the actual and forecasted cash for each year in the post-crisis period is taken to account for abnormal cash. A simple hypotheses test for a mean different than zero is used to test the statistical significance of the forecasting errors. The same is performed for the base period and each year of the pre-crisis period, meaning that abnormal cash in the pre-crisis period is measured as the difference between actual cash and forecasted cash from a model based on the base period. The results of this thesis are based on models estimated over longer periods than the three-year period selected in Pinkowitz et al. (2013) and are not given as periodical averages, but on a per-year forecasting error basis. For unbiased results both whole sample and regional specific forecasting models are estimated.
4. RESULTS AND DISCUSSION
The results of this thesis are divided in two tables. Table VI reports the results of the OLS regressions testing for the relationship between firm characteristics and cash, the differences of this relationship between different regions and the change, if any, after the financial crisis. Table VII presents the forecasting errors of the models, which are estimated from the pre-crisis period, for the whole sample and the different regions separately.
Table VII
OLS Regressions Estimating Cash Holding Determinants
All models are OLS regressions with CASH as a dependent variable. Model 1 in Panel A includes all variables from Table II. Model 2 includes lagged cash [CASH(t-1)] to check for
18
Panel A
Model Theory 1 2 3
Dependent Variable CASH CASH CASH
Trade-off Pecking Order CASH(t-1) 0.046 0.006 (12.0)*** (2.0)* VOL + 0.175 0.170 0.009 (24.4)*** (21.6)*** (1.0) MB + + 0.096 0.097 0.059 (103.9)*** (95.0)*** (54.4)*** SIZE - + -0.058 -0.057 -0.361 -(40.4)*** -(36.0)*** -(86.2)*** CF - + -0.221 -0.220 -0.101 -(56.3)*** -(51.7)*** -(23.8)*** NWC - -0.219 -0.227 -0.382 -(28.3)*** -(26.8)*** -(36.5)*** CAPEX + + 1.167 1.205 1.112 (52.8)*** (49.2)*** (42.2)*** LEV +/- - -0.159 -0.175 0.087 -(15.7)*** -(15.7)*** (7.0)*** RD + + 0.034 0.031 0.013 (30.3)*** (26.9)*** (14.6)*** DUMDIV - -0.101 -0.111 0.104 -(16.0)*** -(15.8)*** (10.7)*** Intercept 0.731 0.709 4.335 (41.3)*** (36.5)*** (86.1)*** Adj. R2 0.304 0.303 0.732 F-stat 4510.459 3423.618 12.25782 Sample 1991-2012 1992-2012 1992-2012 Firms 15491 14912 14333 Observations 121314 103061 103061
Year and Firm FE
Yes
19
The results are also consistent with previous literature (Opler et al., 1999; Bates et al., 2009; Ferreira and Vilela, 2004; Pinkowitz et al., 2013) except the result for capital expenditures. All three models return a positive and highly significant coefficient for capital expenditures. Capital expenditures can be seen as a proxy for profitable future investments, just like the market-to-book ratio, and according to both the trade-off theory and the pecking order theory, firms hold more cash because the positive NPV of these investments disappears in case of bankruptcy. None of the previous papers in the cash holding determinants literature has found such a strong positive and statistically significant relationship between cash and capital expenditures. Thus, the second major finding of this thesis is that it provides empirical evidence that firms with higher capital expenditures hold more cash.
The third major finding of this thesis is the significant positive dynamic effects in cash holdings. Firms with more cash in the previous year have more cash in the current year. Previous studies do not test for any dynamic effects, whose results might suffer from significant auto-correlation.
According to a redundancy test, firm and year fixed effects are significant in Model 3. Random effects were proven insignificant with a Hausman test. The year-fixed effects coefficients, as shown in table 1 in appendix D, indicate that the start of the abnormal cash hoarding to have started in 2001, therefore the base period of the sample is defined as the period until 2000, the pre-crisis period is between 2001 and 2007, and the post-crisis period is between 2008 and 2012.
Table VII Panel B
Model 4.
Dependent Variable CASH
20 LEV -0.262 0.250 -0.140 0.160 -0.192 0.030 -(11.9)*** (4.5)*** -(2.5)* (4.7)*** -(2.3)* (0.3) RD 0.015 0.531 0.275 0.137 -0.273 0.050 (11.9)*** (7.9)*** (9.3)*** (24.2)*** -(2.7)*** (0.8) DUMDIV -0.082 0.043 0.211 0.054 -0.002 -0.091 -(4.2)*** (1.2) (6.1)*** (1.9)* (0.1) -(1.7)* Intercept 0.973 -0.599 -0.065 0.298 -0.235 -0.330 (21.0)*** -(5.6)*** -(0.7) (4.0)*** -(1.6) -(2.2)* Adj. R2 0.308 F-stat 335.9 Sample 2001-2012 Firms 11808 Observations 64123
Panel B of Table VI includes the results of the OLS regression with interactions between the dependent variables and dummy variables for the regions before and after the crisis. For easier comprehension of the results, Panel C provides the coefficients of the same regression in absolute terms. One important finding is that firms in the EMU region with more operating cash flows hold more cash before the crisis. This is contradictory to both other regions and to the findings in Panel A. After the crisis, however, the coefficient changes its sign from 0.071 to -0.444. This means that EMU firms started following the trade-off theory view on cash flows only after the crisis, since before that the pecking-order logic prevails. Another finding is that the coefficient for capital expenditures of EMU firms suffers the biggest change, since it more than triples - from 0.524 before the crisis, to 1.776 after the crisis. At the same time, the coefficients of US and UK firms decrease compared to before the crisis period.
A similar contrasting reaction is observed for the coefficients of leverage. The coefficient for leverage of EMU firms decreases sharply from -0.012 to -0.454, while the coefficients of US and UK firms are still negative, but decrease by half after the crisis.
21
Table VII Panel C
One major difference between UK firms and the other two regions is that before the crisis firms that paid out dividends held more cash. After the crisis, however, the coefficient changes signs, although only significant at the 10% level. The coefficients for US and EMU firms are negative before and after the crisis.
To summarize the results from Panels B and C of Table VII, compared to the US, EMU firms differ significantly in their cash holding determinants and their changes from before to after the crisis. On the other hand, UK firms are similar to US firms at least in terms of signs of the coefficients, except for the dividend dummy sign before the crisis.
Table 2 in appendix D reports that, based on a model from 1991 until 2000, there has already been a significant increase in cash holdings in each region in the years between 2001 and 2007. Table VIII reports the results of the forecasting errors of the models created from the cross-sectional regressions for the years between 2001 and 2007.
22
Table VIII
Cash Forecast Errors Based on the 2001 – 2007 period
The table reports the average differences between the actual cash ratios and the cash ratios predicted by an out-of-sample model. The forecasts are estimated from models created by taking the averages from Fama-MacBeth cross-sectional regression coefficients for the period between 2001 and 2007. Region specific models are estimated. The model for the whole sample is Cash = 0.566 + 0.046 Cash(t-1) + 0.249 VOL + 0.078 MB - 0.052 SIZE - 0.058 CF – 0.201
NWC + 1.049 CAPEX - 0.126 LEV + 0.652 RD – 0.063 DUMDIV. The model for the US is Cash = 0.651 + 0.051 Cash(t-1) + 0.363 VOL + 0.074 MB - 0.062 SIZE - 0.018 CF – 0.222 NWC + 1.125
CAPEX - 0.156 LEV + 0.740 RD – 0.056 DUMDIV. The model for the UK is Cash = 0.787 - 0.017
Cash(t-1) + 0.270 VOL + 0.082 MB - 0.089 SIZE - 0.193 CF – 0.199 NWC + 1.070 CAPEX - 0.310
LEV + 0.494 RD + 0.124 DUMDIV. The model for the EMU is Cash = 0.334 + 0.004 Cash(t-1) + 0.131
VOL + 0.047 MB - 0.028 SIZE + 0.106 CF – 0.065 NWC + 0.538 CAPEX - 0.001 LEV + 0.578 RD – 0.035 DUMDIV. t-statistic for a test of mean different than 0 is reported in parentheses.
Year Whole Sample US UK EMU 2008 -0.059 -0.074 -0.083 -0.046 -(0.6) -(3.1)*** -(3.9)*** -(4.2)*** 2009 0.074 0.115 0.043 0.041 (5.2)*** (5.1)*** (1.7) (4.2)*** 2010 0.077 0.064 0.075 0.155 (1.3) (1.2) (3.7)*** (11.5)*** 2011 0.085 0.084 0.042 0.054 (7.1)*** (4.5)*** (2.8)*** (5.7)*** 2012 0.088 0.069 0.073 0.082 (7.9)*** (3.4)*** (4.6)*** (9.8)***
***Significant at 1% (two sided). *Significant at 10% (two sided).
23
5. CONCLUSION
This master’s thesis investigates whether the increase of cash holdings over time and the differences between firms in the US, UK, and EMU regions, can be explained and predicted by fundamentals. The major findings of this master’s thesis confirm and build on previous literature. Most of the results confirm the trade-off concept of corporate structure, with some exceptions. EMU firms with high research and development expenses hold less cash after the crisis. This contradicts the precautionary motive for cash holdings. Before the crisis, the relationship is positive, consistent with all previous empirical evidence. Moreover, after the crisis, UK firms that pay out dividends hold more cash than non-dividend paying firms. One argument is that firms that pay dividends simply ensure that they have enough cash to pay out the next dividends. However, this also contradicts all previous empirical evidence on the matter. EMU firms before the crisis, which had higher operating cash flows held more cash, which is consistent with a pecking-order world. After the crisis however, the behaviour of EMU firms changes, and firms with high operating cash flows hold less cash, which is consistent with previous literature and empirical evidence.
The results of this master’s thesis show significant and positive dynamic effects in cash holdings. Firms that had more cash the previous year, hold even more cash in the current year. This is consistent over all regions in all sample years.
This master’s thesis provides empirical evidence for a similar cash holding behaviour between regions in the years after the crisis. Firms in all regions reacted negatively to the crisis, and in 2008 the increase in cash holdings, which is evident in the pre-crisis years, drops significantly. From 2009 until 2012, the demand for cash increases abnormally, and firms in all regions increase their cash holdings at a higher rate than before the crisis. The motivation for this behaviour is that the precautionary motive for holding cash is excessively strong after the crisis of 2008.
24
REFERENCE LIST
Bates, T. W., Kahle, K. M., and Stulz, R. M., 2009. Why Do U.S. Firms Hold So Much More Cash than They Used To? The Journal of Finance 64, 1985-2021.
Baumol, W. J., 1952. The Transactions Demand for Cash: An Inventory Theoretic Approach. Quarterly Journal of Economics 66, 545-556.
Brooks, C., 2008. Introductory Econometrics for Finance. 2nd edition. Cambridge: Cambridge University Press.
Donaldson, G., 1961. Corporate Debt Capacity: A Study of Corporate Debt Policy and the Determination of Corporate Debt Capacity. Beard Books. Published in 2000.
Fama, E.F., MacBeth, J.D., 1973. Risk, Return, and Equilibrium: Empirical Tests. Journal of Political Economy 81, 607-636.
Frank, M., Goyal, V., 2003. Testing the Pecking Order Theory of Capital Structure. Journal of Financial Economics 67, 217-248.
Ferreira. M. A., and Vilela, A. S., 2004. Why Do Firms Hold Cash? Evidence From EMU Countries. European Financial Management 10, 295-319.
Foley, C., Hartzell, J., Titman, S., Twite, G., 2007. Why do Firms Hold so Much Cash? A Tax-based Explanation. Journal of Financial Economics 86, 579-607.
Hanlon, M., Maydew, E., Saavedra, D., 2014. The Taxman Cometh: Does Tax Uncertainty Affect Corporate Cash Holdings? Unpublished working paper. Electronic copy available at: http://ssrn.com/abstract=2292020
Hirschleifer, J., 1966. Ivestment Decision under Uncertainty: Applications of the State-Preference Approach. Quarterly Journal of Economics 80, 252-277.
Keynes, J.M., 1936. The General Theory of Employment. In: Interest and Money. Harcourt Brace, London.
Kim, C., Mauer, D., Sherman, E., 1998. The Determinants of Corporate Liquidity: Theory and Evidence." Journal of Financial and Quantitative Analysis 33, 335-359.
Modigliani, F., Miller, M., 1958. The Cost of Capital, Corporation Finance and the Theory of Investment. The American Economic Review 48, 261-297.
Modigliani, F., Miller, M., 1961. Dividend Policy, Growth, and the Valuation of Shares. The Journal of Business 34, 411-433.
25
Myers, S., 1984. The Capital Structure Puzzle. Journal of Finance 39, 575-592.
Myers, S., Majluf, N., 1984. Corporate Financing and Investment Decisions when Firms Have Information that Investors do not Have. Journal of Financial Economics 13, 187-221. Opler, T., Pinkowitz, L., Stulz, R., Williamson, R. 1999. The Determinants and Implications
of Corporate Cash Holdings. Journal of Financial Economics 52, 3-46.
Ozkan, A., Ozkan, N., 2004. Corporate Cash Holdings: An Empirical Investigation of UK Companies. Journal of Banking and Finance28, 2103-2134.
Pinkowitz, L.,Williamson, R., 2004. What is a Dollar Worth? The Market Value of Cash Holdings. Unpublished working paper, Georgetown University.
Pinkowitz, L., Stulz, R., Williamson, R., 2013. Is there a US Cash Holding Puzzle after the Financial Crisis?. Unpublished working paper.
Robichek, A., Myers, S., 1966. Problems in the Theory of Optimal Capital Structure. Journal of Financial and Quantitative Analysis 1, 1-35.
Scott, J., 1976. A Theory of Optimal Capital Structure. The Bell Journal of Economics 7, 33-54.
Shyam-Sunder, L., Myers, S.C., 1999. Testing Static Trade-off against Pecking Order Models of Capital Structure. Journal of Financial Economics 51, 219-244..
26
APPENDIX A
Percentage of Negative Net Leverage Firms
This table reports the percentage of firms that have negative net leverage. Net leverage is calculated as leverage net of cash. Leverage is calculated as book value debt to invested capital net of cash and equivalents. Cash is calculated as book value of cash and securities to invested capital net of cash and equivalents. All firm data is downloaded for the period between 1991 and 2012 for firms based in the US, UK, or any country fully participating in stage three of the European Monetary Union as of 1st of March 2014, which includes Belgium, Germany, Estonia, Ireland, Greece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovenia, Slovakia and Finland. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded from the sample. Only firms with positive invested capital and book value of equity are included. Firm-year observations that miss an observation for any datatype are excluded.
27
APPENDIX B
Firm-year Observations per Region
The sample includes firms based in the US, UK, or any country fully participating in stage three of the European Monetary Union as of 1st of March 2014, which includes Belgium, Germany, Estonia, Ireland, Greece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovenia, Slovakia and Finland. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded from the sample. Only firms with positive invested capital and book value of equity are included. Firm-year observations that miss an observation for any datatype are excluded.
28
APPENDIX C.1
Average Cash Difference between R&D paying firms and firms with R&D expenses zero The table reports the difference of the average cash for firms that have positive research and development expenses net of the average cash for firms that report zero research and development expenses. All firm data is downloaded for the period between 1991 and 2012 for firms based in the US, UK, or any country fully participating in stage three of the European Monetary Union as of 1st of March 2014, which includes Belgium, Germany, Estonia, Ireland, Greece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovenia, Slovakia and Finland. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded from the sample. Only firms with positive invested capital and book value of equity are included. Firm-year observations that miss an observation for any datatype are excluded.
R&D payers - R&D non-payers Year
Whole
Sample USA UK EMU
29
APPENDIX C.2
Average Cash of Dividend Payers net of the Average Cash for Non-Dividend Payers The table reports the difference of the average cash for dividend paying firms net of the average cash for non-dividend paying firms. All firm data is downloaded for the period between 1991 and 2012 for firms based in the US, UK, or any country fully participating in stage three of the European Monetary Union as of 1st of March 2014, which includes Belgium, Germany, Estonia, Ireland, Greece, Spain, France, Italy, Cyprus, Luxembourg, Malta, the Netherlands, Austria, Portugal, Slovenia, Slovakia and Finland. Financial firms (SIC codes 6000-6999) and utilities (SIC codes 4900-4999) are excluded from the sample. Only firms with positive invested capital and book value of equity are included. Firm-year observations that miss an observation for any datatype are excluded.
Dividend payers - Non-dividend payers Year
Whole
Sample USA UK EMU
30
APPENDIX D.1
Year Fixed Effects Coefficients of Model 3 from Table VII
The model is a simple panel data OLS model which includes all firm characteristics as independent variables and cash as a dependent variable. The results indicate that the start of the abnormal cash hoarding has started in 2001, therefore the base period of the sample is defined as the period until 2000, the pre-crisis period is between 2001 and 2007, and the post-crisis period is between 2008 and 2012.
31
APPENDIX D.2
Cash Forecasts Errors Based on the 1992 – 2000 period
The table reports the average differences between the actual cash ratios and the cash ratios predicted by an out-of-sample model. The forecasts are estimated from models created by taking the averages from Fama-MacBeth cross-sectional regression coefficients for the period between 1992 and 2000. Region specific models are estimated. The model for the whole sample is Cash = 0.621 + 0.031 Cash(t-1) + 0.247 VOL + 0.068 MB - 0.063 SIZE - 0.002 CF –
0.100 NWC + 0.084 CAPEX + 0.011 LEV + 0.799 RD – 0.033 DUMDIV. The model for the US is Cash = 1.042 + 0.072 Cash(t-1) + 0.334 VOL + 0.065 MB - 0.101 SIZE + 0.004 CF –
0.158 NWC + 1.016 CAPEX - 0.008 LEV + 0.810 RD – 0.039 DUMDIV. The model for the UK is Cash = 0.461 - 0.057 Cash(t-1) + 0.047 VOL + 0.047 MB - 0.049 SIZE + 0.184 CF –
0.137 NWC + 0.239 CAPEX - 0.362 LEV + 0.209 RD + 0.131 DUMDIV. The model for the EMU is Cash = 0.275 + 0.014 Cash(t-1) – 0.001 VOL + 0.039 MB - 0.031 SIZE + 0.402 CF –
0.167 NWC + 0.316 CAPEX + 0.152 LEV + 0.751 RD – 0.043 DUMDIV. t-statistic for a test of mean different than 0 is reported in parentheses.
