University of Amsterdam
Amsterdam Business School
MSc Business Economics, Finance track
Master Thesis
The determinants of cash savings after the financial crisis
Student: Alena Nasedkina 10825231 Supervisor: dr. Rafael P. Ribas
2 Statement of Originality
This document is written by Alena Nasedkina who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3 Abstract
In this thesis I examine the determinants of cash holdings for public U.S. firms in the 2000-2013 period. In this paper normal cash holdings are defined as holdings after the financial crisis that a company with the same characteristics would have had before the crisis. Using cross-sectional and panel data, I find evidence that abnormal cash ratio equals 6% of cash savings after the 2008 financial crisis. This abnormal growth of cash-to-assets ratio is concentrated among dividend non-paying firms, companies that repurchase their stock, and financially constrained firms. In contrast, multinational companies are not significantly different from domestic firms in defining their cash policy despite tax incentives the former get. In sum, high cash holdings of U.S. companies can be explained by a transaction motive, precautionary motive (hedging needs) and indirectly by an agency motive that is presented in stock repurchases.
4
Contents
Statement of Originality ... 2 Abstract ... 3 1. Introduction ... 5 2. Literature Review ... 82.1 Motives for holding cash ... 8
2.2 Empirical literature ... 10
3. Hypotheses and Methodology ... 13
4. Data ... 16
4.1 Data description... 16
4.2 Summary statistics... 19
5. Results ... 27
5.1 Examination of abnormal cash holdings ... 27
5.2 Definition of determinants of abnormal cash holdings ... 29
6. Robustness Checks ... 37
7. Conclusion ... 40
8. References ... 42
5 1. Introduction
Managers often balance between internal and external financing of profitable investment projects. In good times debt financing helps a company to increase the returns on equity, but in downturns high debt may cause troubles. A similar situation occurs with cash holdings. On the one hand, a high level of cash holding can be a signal of strong performance, meaning that cash is accumulated so fast that management cannot find investment opportunities at that moment of time. On the other hand, holding much cash could mean that there are no attractive opportunities for investments or that the management is just reluctant to use the existent opportunities. As a result, high cash holdings can have either positive or negative effects on firm value. Thus it is extremely important to understand the motives for firms’ cash saving behavior.
Over the last decade a number of studies have investigated the determinants of cash holding in the U.S., exploring firms’ characteristics such as market-to-book ratio, size, capital expenditures and R&D expenses. However, after the financial crisis in 2008 fair attention has been paid to extremely high cash savings of U.S. firms (Figure 1). For example, one headline puts it “Fun Number; Apple Has Twice as Much Cash as the US Government”.1
From several determinants that explain why some companies hold more cash than others, can some of them explain that upward trend? Accordingly, this thesis attempts to answer the following questions: 1). Are companies’ fundamentals able to predict their cash savings during and after financial crisis? 2). Which determinants explain the changes in firms’ cash ratio?
Some argue that such abnormally high amounts of cash are due to immoderate uncertainty in capital markets. For instance, Bates, Kahle and Stulz (2009) which found that the average cash-to-assets ratio for U.S. manufacturing companies has increased from 10.5% in 1980 to 23.2% in 2006 argue that the main reason for such a dramatic increase in the last twenty years is cash flow uncertainty, as well as the reduction in capital expenditures. Another explanation
6 could be that multinational companies have tax incentives to keep their money abroad in cash.2 For instance, if a subsidiary of a U.S. company operates in the United Kingdom, it pays England corporate income tax of 21% on its profits. When these profits are back to the USA, an additional tax is paid which equals the difference between the U.S. tax rate (35%) and the UK tax rate (21%). However, this subsidiary does not pay the difference to its government until these profits are brought back to the country. The third explanation is that CEOs hold cash to protect their own assets against stock depreciation. Since executives are paid large amounts in the form of options, they could use cash to avoid negative fluctuations through stock repurchases.
Figure 1. Aggregate Cash and Equivalents of Non-Financial Non-Utility U.S. Firms
SOURCE: Compustat.
This thesis contributes to the existent literature in the following ways. First, I will analyze cash holdings of U.S. firms just before the crisis, unlike earlier researches that used two or even three previous decades for defining the determinants. Second, to the best of my knowledge, abnormal cash savings and their determinants were not investigated before. Pinkowitz, Stuls and Williamson (2013) explore abnormal cash holdings in their paper but do not attempt to explain which firms contributed most.
It is found out that after the financial crisis cash savings of U.S. companies cannot be explained by firms’ characteristics and that abnormal cash holdings amount to 6.4%.
2 How Much Do U.S. Multinational Corporations Pay in Foreign Income Taxes? 18 March,
7 Furthermore, this increase in cash holdings is by dividend non-paying firms, companies that make stock repurchases and financially restricted firms. As every company regardless of its field of activity holds cash and has to decide on its level, the results of this thesis may be useful for firms planning their financial policy.
The research is structured as following. Section 2 is a discussion of the related literature. Namely, different motives for holdings cash are determined, and empirical papers are analyzed according to those motives. Section 3 is dedicated to formulation of hypotheses as well as methodology description. Section 4 reveals what data is used and how it was collected. Moreover, this section also includes summary statistics of the data. Section 5 represents the main results and their interpretation, while Section 6 provides with some robustness checks. Finally, the conclusion describes the answers for the research questions. It also involves limitations and implications of the paper, as well as directions for future research.
8 2. Literature Review
2.1 Motives for holding cash
There is a lot of empirical literature that explains cash holdings by firm characteristics. Nevertheless, the existent literature does not provide a common opinion regarding these factors.
There are two well-known theories about motives of holding cash. According to the pecking order model, there is no optimal cash ratio and cash can be considered as negative debt. According to this model, as a company accumulates funds internally, its debt decreases. When a firm experiences a fall in its internal funds, its cash ratio decreases, and net debt, consequently, increases. In this perspective firms are indifferent between accumulating cash and repaying debt.
An alternative view of holding cash is provided by a trade-off theory which argues that firms balance between costs and benefits of high cash ratio. It implies that there exists the optimal cash ratio that maximizes shareholders’ value. Thus companies should choose it in such a way that the marginal cost of holding cash equals its marginal benefit. The cost of holding more cash than necessary is reflected in the lower rate of return on cash because of the absence of a liquidity premium. This premium is only paid for assets which cannot be converted into cash easily and at the fair value. Thereby cash itself is free of this liquidity premium which makes holding extra cash unprofitable. On the other hand, there are benefits associated with holding cash and liquid marketable securities. The first one is that high cash holdings let save transaction costs to raise funds and not liquidate assets to make payments. Second, a company can use money to finance its investments and other activities. Keynes (1936) describes the first benefit as a transaction cost motive, and the second benefit as a precautionary motive for holding cash. However, despite this trade-off between costs and benefits, managers prefer holding extra cash on a balance sheet. This can be explained by the fact that cash reduces firm risk and managers pay too much attention to the precautionary motive as explored by Gao and Yaniv (2014).
9 As was already mentioned, firms hold cash for a variety of reasons that can be divided into the four categories: the transaction motive, the precautionary motive, the agency motive and the speculative motive.
The transaction motive refers to companies’ operation activities. They include both cash inflow and outflow. On the one hand, a firm is constantly involved in purchasing goods or services that it has to pay for. For instance, it makes payments in terms of cash for paying salaries, rent, interest, taxes, dividends, etc. On the other hand, it receives money as revenues, interests and so on. If these inflows and outflows were perfectly matching, there would be no need for a company to hold cash. However, in reality it is impossible, that is why in times, when costs exceed income, a firm has to hold a certain level of cash on its balance sheet to meet paying requirements. According to Custodio et al. (2005), firms hold more cash especially in times of recessions. During the crisis it is more costly to be short of money as it is difficult to convert assets into liquidity. Moreover, Fama (1991) found out that capital markets are also costly to access because of high debt agency costs and high credit spreads.
The precautionary motive refers to a firm’s unpredictable events and emergencies such as fall in the demand of its products, rising labor and material costs and so on. The cash level for this purpose depends on the degree of predictability on the market and for a certain company. For example, if a firm operates in an unstable environment, or its future cash inflows are difficult to predict, or it will have attractive opportunities in the future, it should hold more cash as a precaution. In contrast, when the economic conditions improve, a company has more chances to liquidate assets or access capital markets in order to get money it needs. However, some firms that are in good standing with banks have strong borrowing capacity and do not keep money as a precautionary motive. Almeida, Campello, Weisbach (2004) in their paper explored this motive for keeping cash and found out that financially restricted companies invest in cash out of their cash flows, while unconstrained companies do not.
10 The agency motive refers to managers’ decisions of paying dividends to their shareholders or keeping money in a firm for future investment opportunities that may arise. However, entrenched management would also retain cash even when the firm does not have attractive investment opportunities. Dittmar et al. (2003) found that companies used to hold more cash in those countries, where the agency problems are more common. Moreover, they proved that in such countries entrenched CEOs usually have excess cash but spend it quickly.
The speculative motive refers to possible favorable opportunities that may arise at any time. Sometimes companies hold cash in excess to transaction and precautionary needs because they have speculative needs. For example, a company knows that in the next period prices for raw materials will substantially decrease for some time. In this case it will hold cash in order to have an opportunity to buy more raw materials at a low cost. Another possibility is purchasing securities once a firm found out that interest rates will grow.
2.2 Empirical literature
A paper by Opler, Pinkowitz, Stulz, Williamson (1998) was one of the first that addressed this issue exploring cash holdings determinants and implications for public firms from 1971 to 1994. They found that companies with considerable growth opportunities and riskier cash flows hold higher ratios of cash to total assets. On the contrary, firms with more access to the capital markets – i.e. large companies and companies with good credit ratings - tend to hold less cash.
More recently, Almeida, Campello, Weisbach (2004) study the link between corporate liquidity demand and firms’ financial constraints. For that purpose they use the cash flow sensitivity of cash to test for financial constraints. U.S. manufacturing firms over the 1970 to 2000 period are classified as financially (un)constrained according to five different approaches: firm size, payout policy, bond ratings, commercial paper ratings, and KZ index3. They find that the level of cash holdings takes into account the relative attractiveness of current and future investments. As predicted, financially constrained firms have a much higher cash flow sensitivity
3
11 than unconstrained firms, which reveals that imperfect capital markets are the reason for holding extra cash. In their paper a change in cash holdings is used as a dependent variable, while a firm cash flow is an explanatory variable. The model includes controls for size and growth opportunities approximated by market-to-book ratio.
Yet not every financially constrained firm presents this cash flow sensitivity of cash. According to Acharya, Almeida, Campello (2006), financially constrained firms and companies with high hedging needs hold cash rather than reducing outstanding debt, while other firms either have more access to capital markets and decrease their debt in order to “save borrowing capacity” or have low hedging needs and do not have to keep much cash. Their findings are in accordance with Bates, Kahle, Stulz (2009) who point out that the precautionary motive is a critical determinant for the dramatic increase in cash holdings. Despite the market for derivatives has been growing for the last decade, U.S. firms either cannot hedge or are simply reluctant to use derivative instruments. Instead, they hold their profits in the form of cash. Another finding is that the increase in cash holdings is mostly driven by firms that do not pay dividends on a regular basis. In particular, they show that the average cash ratio of dividend payers is almost the same in 2004 as in 1980. In contrast, the mean and the median cash ratios of non-dividend payers increased by 112% and 224%, respectively.
An additional explanation for increasing cash holdings is the tax incentives that multinationals have in holding their profits in cash abroad. This issue is raised by Foley et al. (2006), who argue that in the U.S. firms that have foreign operations are provided with tax credits for foreign income taxes paid abroad. This creates incentives for multinationals to keep earnings abroad in the form of cash if they do not have attractive projects to invest in. Moreover, the sensitivity of cash holdings to repatriation taxes is particularly high for technology-intensive companies, but there is no evidence that it is true for financially constrained firms.
A recent paper by Pinkowitz, Stuls, Williamson (2013) investigates cash holdings after the financial crisis in 2008. They also argue that the substantial increase is driven by
12 multinationals and more profitable firms. They estimate a model of cash ratios for 1980-2000 and analyze abnormally high cash after 2000. However, there is a problem of choosing a sample period in the paper. The point is that since 1980 a large progress in business has been observed and firms’ characteristics have also changed dramatically, and using data from 1980 could possibly bias the estimations.
However, Liu and Mauer (2010) provide with another perspective to look at cash holdings. They examine the relationship between cash savings and CEO compensation schemes. They find that there is a positive relationship between risk-taking incentives and cash holdings which means that CEOs can partly regulate the supply of shares and avoid stock depreciation by repurchasing them. Therefore, either the frequency of stock repurchases or the percentage of compensation paid in stock options (or the combination of both) can also be a determinant of cash savings.
13 3. Hypotheses and Methodology
In most studies, mentioned above, high cash holdings are related to firms’ fundamentals (except for CEO option compensation and stock repurchases). If this is the case, the substantial increase in cash ratios should be explained by changes in these fundamentals over the years. Another hypothesis is that the fundamentals have not changed as much as the cash sensitivity to these characteristics. That is, the coefficient of explanatory variables (firm characteristics) on cash savings has changed. Finally, the increase in cash ratios might have increased due to an unexplained component (or unobserved characteristics). Given these three components I investigate which types of firms - in terms of tax benefits, precautionary motives, CEOs’ protection and dividends – have driven the upward trend in cash.
First, I test whether firms’ cash holdings behavior during and after the financial crisis has been significantly and fundamentally different from their behavior before the crisis. As long as cash sensitivity to firm characteristics has changed, the pre-crisis model cannot predict existing cash ratios. Then I verify whether this change has happened only in particular firms, namely: multinationals (tax benefits motive), financially constrained firms (precautionary motive), non-dividend payers, and firms in which CEOs are encouraged to protect themselves from stock depreciation.
Thus, the hypotheses tested in this thesis are the following:
Hypothesis 1. U.S. firms hold abnormally high cash after the financial crisis, so that fundamentals do not entirely explain the growth in cash savings (based on the pre-crisis model). Hypothesis 2. Growth in abnormal cash holdings is higher in:
3.1 Financially constrained firms 3.2 Firms that do not pay dividends 3.3 Multinationals
3.4 Firms protecting themselves against stock depreciation.
To test the first hypothesis, ordinary least squares regression models are used. It will be tested using the following model. The dependent and independent variables are listed in Table 1.
14
, (1)
Once the pre-crisis coefficients are estimated, I use them to predict cash ratios in the crisis and post-crisis periods. Further, actual cash savings are compared with the predicted ones. This procedure let reveal abnormally high or low cash ratios of manufacturing firms. The significance of this difference is tested through a simple two-population mean test.
To test hypothesis 2, the sample is divided into two groups for each criterion (eight subsamples in total).They are: constrained and unconstrained companies, multinationals and domestic firms, dividend payers and non-dividend payers, firms that repurchase their stock and that do not. Then equation (1) is estimated for each subsample.
The difference in cash holdings between groups over time is analyzed using the approach developed by Juhn, Murphy and Pierce (1993), henceforth JMP. The rise in cash holdings can be explained by both observed fundamentals such as size, leverage, cash flow and capital expenditures, and unobservable characteristics (residuals). JMP suggest a useful method for isolating these effects. Consider the simplified version of equation (1), as follows:
, (2) where Yit is the cash ratio for a company i in year t, Xit is a vector of individual characteristics, and uit is the component of cash ratios accounted for by the unobservables. In other words, uit has two components: an individual’s percentile in the residual distribution, Өit, and the distribution function of the cash ratio residuals, Ft ( ). Thus, residuals are:
( | ) (3) where ( |
) is the cumulative residual distribution for firms with characteristics in year t.
In this framework changes in the distribution of cash ratios come from three different sources: changes in the distribution of fundamentals (X), changes in the cash sensitivity to these individual characteristics (β) and changes in the distribution of the residuals (u).
15 The first step is to calculate cash ratios with fixed observable coefficients and a fixed residual distribution:
̅ ̅̅̅̅̅( | ) (4) Second, I allow observable quantities and coefficients to change over time, so the
counterfactual cash ratios are the following:
̅̅̅̅̅( | ) (5) Finally, I allow observable quantities, coefficients and distribution of residuals to vary:
( | ) (6) This technique is used to calculate the counterfactual cash ratios (Y1it, Y2it , and Y3it)for each period (before, during, and after the financial crisis) and group. For each group:
(7) represents changes in cash ratio between period 0 and 1 that are driven by fundamentals,
(8) represents changes in cash ratio that are driven by changes in sensitivity, and
(9) represents changes in cash ratio that are driven by changes in unobservables.
16 4. Data
4.1 Data description
All data come from COMPUSTAT from 2000 to 2013 and only includes firms that satisfy the following restrictions:
Market capitalization is greater than $5 million Annual Sales is greater than $1 million
Moreover, financial firms (SIC code 6000-6999) and utilities (SIC code 4900-4999) are also excluded from the sample because the former meet capital requirements and the latter use cash that is subject to regulatory supervision in some jurisdictions.
Table 1 displays all variables included in the regression model, respective formulas for the variables and expected relationship with a dependent variable (Cash Ratio).
Market-to-book ratio is used as a proxy for investment opportunities for a firm (see Smith and Watts, 1992). Since the costs of financial distress are higher for firms with high market-to-book ratio, such companies hold more cash on their balance sheets. Moreover, capital expenditures and R&D expenses also proxy variables for investment opportunities and/or costs of financial distress. Hence, one would expect a positive relationship between Capex, as well as R&D expenses, and cash ratio. On the other hand, as capital expenditures demand cash to make them, one would expect Capex to reduce cash.
One view is that companies hold liquid assets to make sure they will be able to make payments and invest in future. From this perspective, large firms usually have greater excess to capital markets than small firms, which is why they need not hold so much cash for a precautionary motive. One can argue that financially constrained firms tend to invest in cash out of cash flow as a precautionary motive. In other words companies may hedge themselves from cash deficit in bad states of economy. Furthermore, industry volatility, measured as the average cash flow variance within the industry, influences companies’ cash ratio. Firms prefer to keep more cash as a precautionary motive if they operate in unstable, cyclical and more volatile sector.
17 Non-cash working capital is considered a measure of liquid assets in a firm. Consequently, the more liquid assets a firm has on its balance sheet, the less cash is needed to finance its future investments and current operations.
Due to the agency costs of debt (shareholders’ and debt holders’ interests differ), highly levered companies find it extremely expensive to raise additional funds and therefore, hold more cash to finance its activities. In contrast, some theories explain an additional amount of cash as a negative debt. It means that cash can be easily used to pay debt back, so these variables are expected to be negatively correlated.
Dividend payers should hold less cash than non-dividend payers for several reasons. First of all, dividends paid consume cash (large portion of net income). Moreover, firms that pay dividends are more likely to be mature, stable and not so risky companies which do not have to keep much cash.
Acquisitions consume a lot of cash, so firms that pay for M&A have less cash on their balance sheet. Thus a negative relationship is expected between cash holdings of a firm in year t and its’ acquisitions expenses. As M&A and R&D expenses are zero for many firm-years, these variables are included in the models summing with capital expenditures - i.e. controlling for Capex, M&A and R&D are treated as additional investments. Moreover, two dummies are included, which are equal to one if the firm performs M&A (and R&D expenses) and zero otherwise.
Given the distinct reasons for saving cash, the initial sample is divided into several subsamples. The following categorizations of variables are introduced: multinational and domestic; constrained and unconstrained; dividend-payers and non-payers; CEO mostly compensated in stocks/options. Table 2 presents all categories and explains the assignment criteria.
18 Table 1. Description of Variables
The table displays variables that are used in the model. The data is collected from Compustat database. The first column shows the name of the variable of interest. The second column present the formula used for calculating the variables. There is an expected relation between the variables of interest and the dependent variable in Column (3). Column (4) is intended to demonstrate how the items are called in Compustat database.
Variable (1) Formula (2) Expected relation with a dependent variable (+/-) (3) Compustat items (4)
Cash Ratio Cash over total assets no che/at
Market-to-book ratio
(Book value of assets - Book value of equity + Market value of equity)/Book value of assets
+
(( at- ceq)+( csho* prcc_f))/ at
Size Log of total assets - ln(at)
CF (EBITDA – taxes – interest - common
dividends)/Net Assets -
(oibdp- txt- xint- dvc)/(at-che)
NWC (Working Capital – Cash)/Assets - ( wcap- che)/at
Capex Log of capital expenditures over lagged
assets +/-
ln(capx/l.at)
Leverage Short- and long- term debt over lagged
assets -
( dltt+ dlc)/at
R&D Log of (R&D expenses+Capex) over lagged
assets + ln(capx+xrd)/at
M&A Log of (Acquisitions+Capex) over lagged
assets -
ln(aqc+capx)/at
Volatility Mean of firm variance of CF per industry + no
Dividends 1 if the company i paid dividends in year t
and 0 otherwise - no
Table 2. Description of companies’ categories
This table displays types of companies that are explored in the hypotheses. These firms are compared to domestic companies, financially unconstrained companies, dividend non-paying firms and firms with no stock repurchases respectively.
Type of
companies Description
Compustat item
Multinationals These companies have non-zero balance sheet item “Pretax income-Foreign”. It
means that such firms get income from foreign operations and can be categorized as multinationals.
“pifo”
Constrained firms These are companies which are in the bottom three deciles of the sample ranked
by asset size (from min to max)
“at”
Dividend payers A firm is considered as a dividend payer in year t if it pays dividends (the
payout ratio does not matter) in that year
“dvc”
19
4.2 Summary statistics
According to Table 3, for all manufacturing companies the mean cash ratio is around 19.1% in the pre-crisis period, while the median is only 10.26%. It means that there are several large values in the sample which skew the mean upwards. The average cash ratio increases by 0.63 p.p. and 1.76 p.p. during and after the crisis respectively.
Appendix 1 shows the mean, median and weighted cash ratios (CRs) over the years. Median CR as well as weighted CR grew substantially over time. Namely, the median cash ratio increased by 55% during the pre-crisis period (2000-2006 years) and had its peak of 14.03% in 2010, while weighted cash rose by 43% over the sample. The average cash ratio was the largest in 2013 (21.27% of the total assets). Figure 2 presents the mean, median and weighted cash ratio over the sample. All of them grow steadily in the pre-crisis period and fall sharply during the downturn (in 2008).However, right after the crisis cash holdings increase again and are much higher than in 2006-2007 years.
Figure 3 gives information of cash holdings of all companies in the sample for each year. It is observable that cumulative cash holdings more than tripled from 2000 to 2013 reaching almost 3 billion dollars that year.
Table3. Descriptive statistics of cash ratios by sub-periods
The table shows the cash to assets ratios of U.S. companies. Mean CR is the mean cash to assets ratio in the sample. Median CR is the median cash to assets ratio in the sample. As for the sub-periods, pre-crisis represents 2000-2006, crisis represents 2008-2009,
post-crisis represents 2010-2013.
Sub-period Mean CR Median CR
pre-crisis 19.10% 10.26%
crisis 19.73% 11.93%
20 Figure 2. Mean, median and weighted cash ratios (CR) of
manufacturing U.S. companies
Figure 3. Aggregate cash holdings of manufacturing U.S. firms
Table 4 presents mean and standard deviation of the dependent and explanatory variables. There is a large variation in cash holdings among U.S. manufacturing firms. For instance, the average cash ratio in the base period is 19.1% with a standard deviation of 21.5%.
There is a large variation in cash holdings of US companies through the whole sample period (standard deviation is about 21.5% with a mean value of 19.8% of total assets). In the base period R&D expenses were made by 49% companies of the sample but this number decreased to 46% during the crisis. Furthermore, during 2008-2009 years the number of firms which made M&A decreased comparing to the pre-crisis period, while number of dividend payers increased by 3.9 p.p.. To sum up, despite U.S. companies experienced difficulties during the crisis, cash ratio increased which means that firms preferred to borrow money, cut expenses, and keep liquid assets as cash and short-term investments. Detailed descriptive statistics for 3 sub-periods is submitted in Appendix 2.
0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 2000 2002 2004 2006 2008 2010 2012 cash r atio Mean CR Median CR Weighted CR 0 1000 2000 3000 4000 2000 2002 2004 2006 2008 2010 2012 cash ( b ill io n s)
21 Table 4. Summary statistics
This table displays summary statistics for the dependent variable (Cash Ratio) as well as for the independent variables for three subsamples indicating the base period (2000-2006), the crisis period (2008-2009) and the post-crisis period (2010-2013) and for the whole sample (2013). For the purposes of saving space, this table provides the information only about the mean and the standard deviation of each variable. In order to reduce the potential impact of outliers, each variable is winsorized at the 1% level in each tail of the distribution. The minimum value, maximum value and the number of observations are given in Appendix 2.
2000-2006 2008-2009 2010-2013 2000-2013
Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Mean Std. Dev.
Cash Ratio 0.191 0.215 0.197 0.205 0.208 0.218 0.198 0.215 Market-to-book ratio 2.272 2.039 1.829 1.723 2.241 2.256 2.198 2.066 Size 5.982 2.355 6.241 2.330 6.262 2.397 6.129 2.370 Cash Flow -0.042 0.515 -0.071 0.667 -0.125 0.880 -0.075 0.664 NWC 0.040 0.209 0.019 0.227 -0.001 0.273 0.023 0.227 Capex -3.247 1.125 -3.503 1.233 -3.426 1.245 -3.346 1.192 Leverage 0.241 0.245 0.243 0.264 0.251 0.277 0.245 0.257 R&D -2.631 1.137 -2.790 1.200 -2.694 1.231 -2.672 1.180 M&A -2.930 1.227 -3.201 1.287 -3.096 1.328 -3.021 1.283 Div dummy 0.380 0.485 0.419 0.493 0.404 0.491 0.398 0.490 R&D dummy 0.488 0.500 0.461 0.499 0.482 0.500 0.480 0.500 M&A dummy 0.367 0.482 0.343 0.475 0.344 0.475 0.356 0.479
Table 5. Summary statistics of cash ratio for various types of companies
This table presents summary statistics for cash ratio of companies of interest mentioned in the first column. The estimation period is from 2000 to 2013. Summary statistics includes the mean cash ratio and its standard deviation, minimum and maximum values and the number of observations.
Type Mean Std. dev Min Max Obs
Constrained 0.292 0.262 0 0.895 13316 Unconstrained 0.111 0.116 0 0.871 13323 Multinational 0.203 0.232 0 0.895 25952 Domestic 0.192 0.192 0 0.893 18457 Dividend payers 0.129 0.146 0 0.793 15991 Dividend non-payers 0.238 0.239 0 0.895 28418 Stock repurchases 0.198 0.209 0 0.857 19944 No stock repurchases 0.199 0.222 0 0.895 24465
22 4.2.1 Constrained and unconstrained companies
Figure 4. Mean cash ratio for constrained and unconstrained firms
Figure 5. Mean indexed cash ratio for constrained and unconstrained firm
Table 5 Panel A provides descriptive statistics for financially constrained and unconstrained U.S. companies. The mean cash ratio for restricted firms is more than two times higher than for unrestricted ones. It supports the hypothesis that financially constrained firms tend to hold more cash as a precautionary motive as they do not have much access to the capital markets. According to Figure 4, for the whole sample period constrained companies had higher cash ratio than unconstrained firms did, however both types followed the same trend. In order to ease the comparison, cash ratios in Figure 5 are indexed to an average of 100 in 2000 for these
5% 10% 15% 20% 25% 30% 35% 2000 2002 2004 2006 2008 2010 2012 cash r atio
Mean Constr Unconstr
90 100 110 120 130 140 150 2000 2002 2004 2006 2008 2010 2012 in d e xed c ash r atio
23 three series. It is seen from the graph that from 2000 to 2005 cash ratio for unconstrained firms increased dramatically but began to fall sharply till 2008. However, right after the crisis cash ratio for these companies again rose significantly by 2.7 p.p. in one year. Difference of cash holdings between financially restricted and unrestricted firms is shown in Appendix 3. It is seen from the graph that inequality decreases in 2001-2003 years but then it experiences a substantial rise to 21%.
4.2.2 Multinational companies and purely domestic firms
Figure 6. Mean cash ratio for multinationals and domestic firms
Figure 7. Mean indexed cash ratio for multinationals and domestic firms 14% 16% 18% 20% 22% 2000 2002 2004 2006 2008 2010 2012 cash r atio
Mean Multinationals Domestic
90 100 110 120 130 140 150 2000 2002 2004 2006 2008 2010 2012 in d e xed c ash r atio
24 According to the hypothesis 2, multinationals tend to have higher cash ratios on average because of the tax incentives they get. In fact, the difference between mean cash ratios of international and domestic firms is just 1.1%, as reflected in Table 5 Panel B. However, it can be seen from Figure 6 that at the beginning of the sample period this difference was great and it was decreasing from 2000 to 2008 and almost totally vanished later on. This was achieved by a high increase in cash holdings of domestic firms, as shown in Figure 7 (almost by 50% by the end of the period). Cash ratio inequality between these two types of companies was 4% in 2000 which was evaporated through the sample and was almost zero by the end.
4.2.3 Dividend payers and dividend non-payers
As shown in Table 5 Panel C, the difference of cash ratio between companies which pay dividends and which do not is quite large through the whole sample period (12.9% against 23.8%). This difference is especially high in the base period and declines after the crisis (Figure 8). Nevertheless, companies which pay dividends increased their cash savings by 40% in the pre-crisis period and in 2010 their cash holdings composed 168% of what they were in 2000. Cash ratio inequality was quite high at the beginning of the base period (13%). The difference started falling till 2004, rose substantially in the next 3 years, noticeably declined during the crisis and increased again after the crisis was finished.
Figure 8. Mean cash ratio for dividend payers and dividend non-payers
5% 10% 15% 20% 25% 30% 2000 2002 2004 2006 2008 2010 2012 cash rati o
25 Figure 9. Mean indexed cash ratio for dividend payers and dividend non-payers
4.2.4 Stock repurchases
According to Table 5 Panel D, the mean cash ratios of firms that did stock repurchases and that did not are almost equal. However, Figure 11 shows that the former experienced a significant growth up to 33.5% in 2009 comparing to 2000. The difference of cash holdings for these types of companies was not stable and did not have any trend during the periods of interest.
Figure 10. Mean cash ratio for firms that repurchased their stock and for those which did not
80 100 120 140 160 180 2000 2002 2004 2006 2008 2010 2012 in d e xed c ash r atio
Mean Div payers Div non-payers
14% 16% 18% 20% 22% 24% 2000 2002 2004 2006 2008 2010 2012
26 Figure 11. Mean indexed cash ratio for firms that repurchased stock
and for those which did not
According to these graphs, cash ratio moistly increased in those companies that have lower level of cash. That is, when comparing firms in a cross-section we can see that constrained companies, multinationals and dividend non-paying firms hold more cash on average. However, the growth in cash holdings over time is explained by unconstrained and domestic companies and dividend-payers. 85.00 100.00 115.00 130.00 145.00 2000 2002 2004 2006 2008 2010 2012
27
5.
Results5.1 Examination of abnormal cash holdings
Table 4 presents regression estimates for the base period (2000-2006). Column (1) displays an ordinary least squares regression model. Column (2) includes industry fixed effects, while Column (3) accounts for both industry and year fixed effects. Despite there is no much difference in estimated coefficients across these three models and R2 does not change, it is still more appropriate to use model (3). Column 3 gives more precise estimations and will be used for predicting cash ratios of U.S. firms during and after the crisis. As for the statistical relationship between a dependent variable and explanatory variables, cash holdings increase with market-to-book ratio and R&D expenses. Cash ratio decreases with a firm size, net working capital, cash flow, capital expenditures, leverage and M&A. These results are similar to those of Bates, Kahle, Stulz (2006) even though the sample periods are very different.
Table 4. Regressions for the base period
The table represents regressions built for the base period only (2000-2006). The dependent variable is cash divided by assets. Column (1) in Table 4 presents an ordinary least square regression. The second column includes industry fixed effects that control for industry systematic differences. Column (3) includes industry and year fixed effects. The variable “Volatility” is excluded from the models (2) and (3) because it cannot be estimated together with industry fixed effects. Every model includes clustered standard errors.*, **, *** indicate that the coefficient is significantly different from zero at the 10, 5 and 1% level, respectively.
Variable (1) (2) (3) Q 0.001*** (0.000) 0.001*** (0.000) 0.001*** (0.000) Size -0.005*** (0.001) -0.003*** (0.001) -0.006*** (0.001) Cash Flow -0.016*** (0.001) -0.016*** (0.001) -0.016*** (0.001) NWC -0.024*** (0.004) -0.024*** (0.004) -0.021*** (0.004) Capex -0.019*** (0.002) -0.014*** (0.002) -0.011*** (0.002) Leverage -0.077*** (0.004) -0.073*** (0.004) -0.070*** (0.004) R&D 0.042*** (0.002) 0.038*** (0.002) 0.037*** (0.002) M&A -0.013*** (0.001) -0.013*** (0.001) -0.014*** (0.001) Div dummy -0.014*** (0.003) -0.011*** (0.004) -0.013*** (0.004) R&D dummy 0.040*** (0.005) 0.019*** (0.006) 0.018*** (0.006) M&A dummy -0.016*** (0.002) -0.017*** (0.002) -0.016*** (0.002) Constant 0.233*** (0.009) 0.199*** (0.056) 0.206*** (0.056)
28
Volatility 0.004***
(0.000)
- -
Industry FE No Yes Yes
Year FE No No Yes
Adj R2 0.37 0.37 0.37
On the basis of the model (3) the estimated cash ratio is generated using the following formula:
This formula is applicable for the crisis (2008-2009) period and post-crisis (2010-2013) period. After the estimated cash ratios are generated, two-sample t-test is conducted in order to check whether the estimated ratio is different from the actual one. In Table 5, the estimates of abnormal cash savings are presented. Panel A compares cash savings for the crisis and post-crisis periods with the baseline. According to the results, during the 2008 financial crisis abnormal cash holdings were 5.4%, and they were even higher after the crisis (6.4%) comparing to the base period.
Panel B shows the yearly results. The average abnormal cash holdings increase during and right after the crisis. As was demonstrated in Figure 2, raw cash savings decreased in 2008-2009, while abnormal cash holdings grew from 4.4% in 2008 to 6.4% in 2009. In other words, cash holdings in that period should have fallen even more than they did. After 2010 abnormal cash holdings fell slightly and continued to drop till 2013.
Both tables show that indeed there is a significant positive difference in predicted and actual cash ratios which means that there are abnormally high cash holdings of US firms during and after the 2008 financial crisis. Thus, Hypothesis 1 is proved at 1% significance level.
29 Table 5. Examination of abnormal cash savings
“Difference” is the difference between actual cash holdings of the period and cash savings predicted by the model from Table 4 column (3). .*, **, *** indicate the mean is significantly smaller (larger) than in the base period at the 10, 5 and 1% level, respectively.
Panel A. Examination of abnormal cash holdings by sub-periods
Period Abnormal Difference Mean cash ratios Actual difference in cash ratios
Baseline period 0.000 19.10% 0.000
Crisis period 0.054*** 19.73% 0.033***
Post-crisis period 0.064*** 20.86% 0.092***
Panel B. Examination of abnormal cash holdings by year
Year 2008 2009 2010 2011 2012 2013
Difference 0.044*** 0.064*** 0.070*** 0.063*** 0.057*** 0.067***
Observations 2632 2724 2834 2933 3032 3226
The proof of the hypothesis explicitly demonstrates that the fundamentals did not change as much as the cash sensitivity to US companies’ characteristics changed, and the model cannot perfectly predict existing cash ratios.
Further research is directed to detection of particular types of companies by which these abnormal cash holdings took place. Namely, the following types of companies are explored:
Multinational companies Constrained firms
Dividend payers
Companies with a large amount of stock repurchases
5.2 Definition of determinants of abnormal cash holdings
In order to point out the difference in cash ratios for various types of companies, eight different regression models are built. A comparative analysis of fundamentals’ impact on cash ratio is done for four pairs of companies. The period of interest is from 2007 to 2013 because these types of companies are expected to explain abnormally high cash savings during and after 2008 financial crisis.
Such variables as leverage, R&D expenses, M&A and industry volatility influence cash ratio in the same directions for restricted and unrestricted companies. However, for the former several fundamental variables are insignificant and have no impact on this group’s cash holdings.
30 It is reflected in the adjusted R2 which says that the model explains only 33% of variation in cash holdings for constrained companies, whereas it explains 46% of cash ratio variation for unrestricted firms. The explanation of this difference could be that constrained firms hold cash primarily because of the lack of access to the capital markets, thus firms’ characteristics do not influence cash holdings as much as they do for financially unconstrained firms.
Furthermore, firm size influences cash savings differently for these firms. Large financially constrained companies keep, on average, more cash; while large unconstrained companies, on the contrary, have lower levels of cash savings. At last, crisis has no impact on cash holdings of restricted firms, but it increases cash ratio of unconstrained companies by almost 2% of cash over total assets.
Market-to-book ratio was chosen to proxy investment opportunities because for companies with high market-to-book ratio the costs of financial distress are higher so such firms are expected to hold more cash. However, it has a negative impact on cash for multinational companies. This phenomenon can be explained by another theory mentioned in paper by Stulz (1990) who argues that agency costs are higher for low market-to-book firms and one would expect such firms with entrenched management to keep more liquid assets on their balance sheet. On the other hand, multinationals are such firms with high agency costs because they are costly and more difficult to be managed. That is why US companies that have operations abroad hold less cash when their market-to-book ratio is high.
As for the crisis years, domestic companies increased their cash holdings in 2008 and 2009 by 1.2% on average comparing to the rest of the sample. This could be due to the lack of investment opportunities during this period or fall of total assets. At the same time crisis had no impact on multinational companies’ cash savings as they tend to keep money abroad and are less influenced by economic changes in the USA in contrast to purely domestic firms.
31 Dividend non-payers are freer to choose the level of their cash ratio because they do not have to pay to their investors regularly and in case of absence of profitable investment opportunities can simply hold their retained earnings in cash and marketable securities.
Such fundamental variables as firm size, market-to-book ratio and capital expenditures do not impact the level of cash ratio for companies that do not pay dividends. Furthermore, their amount of cash is no influenced by the crisis, whereas cash savings of dividend paying companies was 1.2% higher during the crisis.
Market-to-book ratio again has a negative impact on cash ratio among companies that do not make stock repurchases and do not have to accumulate cash for these purposes. It could also be due to high agency costs of these companies.
As in previous cases, during the crisis years cash ratio increases but only for companies with stock repurchases and crisis does not have any influence on cash savings for firms that do not.
Table 7 represents cross-sections for crisis and after-crisis years including dummy variables that reflect defined types of companies. It is important to mention that constrained dummy and unconstrained dummy are both included as they are not collinear with each other. First of all, it should be mentioned that size is not significant for any year of the sample so it does not influence cash ratio at all in this cases. Moreover, the coefficient of multinational dummy is not significantly different from zero for any year which means that multinationals in fact are not different from purely domestic firms in defining their policy of cash savings. In contrast, repurchase dummy is highly significant for every year with positive sign. It means that companies that repurchased their stock in year t, had 1.5-2% higher cash ratios than firms that did not. Dividend-paying companies had significantly lower cash savings on their balance sheet than non-dividend-paying companies, especially in 2008 (3.6% lower) and in 2013 (3.5% lower). Constrained companies do not differ from the rest of the sample, while for unconstrained firms
32 cash ratios are significantly lower which proves that such companies do not tend to keep much money because they have more access to the capital markets and can borrow from them.
Thus, hypotheses are partly proved. Namely, stock repurchases and dividends play a significant role in defining the level of cash savings, while multinationals and financially constrained companies do not how any difference from the rest of the sample.
33 Table 6. Panel regressions for various firm types
The table displays panel regressions for each firm type. The sample includes firm-year observations from 2007 to 2013 year. Industry fixed effects, year fixed effects and clustered standard errors are included in the models. The dependent variable in all the models is cash divided by assets. *, **, *** indicate that the coefficient is significantly different from zero at the 10, 5 and 1% level, respectively.
Variable Constrained Unconstrained Multinational Domestic Div. payers Div.
non-payers
Stock Repurchase No Stock Repurchase
Market-to-book ratio 0.000 (0.000) 0.016*** (0.001) -0.001*** (0.000) 0.010*** (0.001) 0.012*** (0.001) 0.000 (0.000) 0.003*** (0.001) -0.001* (0.000) Size 0.039*** (0.004) -0.010*** (0.002) -0.002* (0.001) -0.010*** (0.002) -0.016*** (0.001) -0.002 (0.001) -0.013*** (0.001) -0.006*** (0.001) Cash Flow -0.015*** (0.001) 0.110*** (0.010) -0.013*** (0.001) -0.015*** (0.003) -0.009*** (0.003) -0.014*** (0.001) -0.013*** (0.001) -0.013*** (0.001) NWC -0.001 (0.001) -0.152*** (0.010) -0.002*** (0.001) -0.030*** (0.006) -0.184*** (0.01) -0.003*** (0.001) -0.023*** (0.006) -0.003*** (0.001) Capex -0.006 (0.004) -0.031*** (0.004) -0.005* (0.003) -0.013*** (0.003) -0.030*** (0.004) -0.001 (0.002) -0.022*** (0.003) -0.010*** (0.003) Leverage -0.022*** (0.004) -0.044*** (0.007) -0.027*** (0.003) -0.091*** (0.006) -0.087*** (0.007) -0.034*** (0.003) -0.108*** (0.007) -0.032*** (0.003) R&D 0.023*** (0.003) 0.040*** (0.004) 0.019*** (0.002) 0.035*** (0.003) 0.029*** (0.004) 0.026*** (0.002) 0.037*** (0.003) 0.029*** (0.002) M&A -0.008** (0.004) -0.011*** (0.001) -0.009*** (0.002) -0.014*** (0.001) -0.009*** (0.001) -0.013*** (0.002) -0.014*** (0.001) -0.009*** (0.002) R&D dummy 0.019* (0.011) 0.013*** (0.005) 0.027*** (0.007) 0.025*** (0.007) 0.001 (0.005) 0.046*** (0.007) 0.007 (0.006) 0.035*** (0.006) M&A dummy -0.043*** (0.008) -0.005*** (0.002) -0.021*** (0.003) -0.008*** (0.002) -0.006*** (0.002) -0.022*** (0.003) -0.009*** (0.002) -0.026*** (0.003) Volatility 0.004*** (0.001) 0.001*** (0.000) 0.003*** (0.001) 0.004*** (0.001) 0.001 (0.001) 0.004*** (0.000) 0.003*** (0.000) 0.004*** (0.000) Crisis dummy -0.004 (0.006) 0.019*** (0.002) -0.003 (0.003) 0.012*** (0.003) 0.012*** (0.002) -0.003 (0.003) 0.008*** (0.003) -0.001 (0.004) Constant 0.201** (0.080) 0.12*** (0.041) 0.245*** (0.046) 0.174 (0.137) 0.221*** (0.047) 0.278*** (0.049) 0.322*** (0.048) 0.260*** (0.046) Adj R2 0.33 0.46 0.41 0.40 0.40 0.38 0.43 0.39
34 Table 7. Cross-section regressions for 2008-2013 years
The table represents cross-section regressions for each year from 2008 to 2013. Industry fixed effects, year fixed effects and clustered standard errors are included in the models. The dependent variable in all the models is cash divided by assets. *, **, *** indicate that the coefficient is significantly different from zero at the 10, 5 and 1% level.
Variable 2008 2009 2010 2011 2012 2013 Market-to-book ratio 0.001 (0.001) 0.005*** (0.001) 0.005*** (0.001) 0.005*** (0.001) 0.002*** (0.001) 0.001 (0.001) Size -0.003 (0.003) -0.001 (0.003) -0.004 (0.003) 0.001 (0.003) -0.003 (0.003) -0.005 (0.003) Cash Flow -0.026*** (0.004) -0.009*** (0.003) -0.024*** (0.003) -0.009*** (0.002) -0.024*** (0.003) -0.014*** (0.002) NWC -0.030*** (0.002) -0.004 (0.008) -0.027*** (0.006) -0.030*** (0.007) 0.004** (0.002) -0.003 (0.004) Capex -0.065*** (0.006) -0.049*** (0.006) -0.048*** (0.006) -0.056*** (0.006) -0.046*** (0.006) -0.057*** (0.006) Leverage -0.194*** (0.011) -0.134*** (0.011) -0.127*** (0.010) -0.095*** (0.008) -0.072*** (0.006) -0.093*** (0.009) R&D 0.073*** (0.004) 0.068*** (0.004) 0.076*** (0.004) 0.080*** (0.004) 0.077*** (0.004) 0.086*** (0.004) M&A -0.010** (0.005) -0.015*** (0.006) -0.021*** (0.005) -0.020*** (0.005) -0.028*** (0.005) -0.013** (0.005) R&D dummy 0.001 (0.008) 0.006 (0.008) 0.007 (0.008) 0.015** (0.008) 0.014* (0.007) 0.021*** (0.008) M&A dummy -0.027*** (0.008) -0.020** (0.008) -0.026*** (0.008) -0.025*** (0.008) -0.014* (0.007) -0.028*** (0.008) Volatility 0.001*** (0.000) 0.002*** (0.000) 0.002*** (0.000) 0.002*** (0.000) 0.002*** (0.000) 0.002*** (0.000) Multinational dummy 0.010 (0.006) 0.002 (0.007) -0.002 (0.006) -0.001 (0.006) -0.006 (0.006) -0.005 (0.007) Repurchase dummy 0.020*** (0.006) 0.021*** (0.006) 0.015** (0.006) 0.018*** (0.006) 0.020*** (0.006) 0.016** (0.007) Dividend dummy -0.036*** (0.007) -0.024*** (0.007) -0.022*** (0.007) -0.021*** (0.007) -0.014** (0.007) -0.035*** (0.007) Unconstrained dummy -0.030*** (0.011) -0.038*** (0.012) -0.034*** (0.011) -0.051*** (0.011) -0.039*** (0.010) -0.038*** (0.011) Constrained dummy -0.013 (0.011) 0.004 (0.012) -0.019* (0.011) -0.002 (0.011) -0.008 (0.011) -0.018 (0.012) Constant 0.208*** (0.023) 0.195*** (0.025) 0.244*** (0.023) 0.181*** (0.023) 0.190*** (0.021) 0.266*** (0.023) R2 0.42 0.36 0.42 0.39 0.41 0.45
35 Table 8 estimates the contribution of observed quantities, prices and the unobservables to the increase in differential between corresponding firm types. Panel A shows this difference between domestic and multinational companies through time. According to the results obtained by Juhn-Murphy-Pierce approach, the difference in cash ratios between these types has decreased by 4.4% over the sample. Moreover, this difference fell a lot during the pre-crisis period, then stood unchanged during the crisis and reduced again after it. From 2008 to 2013 the difference in cash ratio declined mainly due to changes in fundamentals, while from 2000 to 2008 it was due to changes in cash sensitivity to these fundamentals. These results support those obtained in Table 5, where it was proved that from 2008 multinationals did not significantly differ from domestic firms in the level of cash ratio.
Panel B represents JMP results for financially constrained and unconstrained companies. During the whole sample period the cash ratio inequality rose by 1.2% (by 0.8% from 2000 to 2010 and by additional 0.4% after the crisis). However, changes in firm characteristics should have decreased this difference by 4.6%, and changes in cash sensitivity should have reduced it even more by additional 11.6%. Nevertheless, the total change increased and it happened due to changes of unobservables. It again confirms the hypothesis that unconstrained firms have decreased their cash savings during and after the crisis.
In order to support the results from Table 5, Panel C gives information about dividend-paying U.S. firms through years. The difference of cash ratios between dividend payers and non-dividend payers increased by 2.9% over the sample. Most of this growth can be attributed to pre-crisis period (2.6% growth). In the pre-pre-crisis period 85% of such a growth is explained by changes in cash sensitivity to companies’ characteristics. More interestingly, after the economic downturn (in 2010-2013) the difference slightly decreased by 0.7% which is equally explained by changes in firms fundamentals, cash sensitivity and changes in unobservables.
According to Panel D, the difference in cash ratios between companies that paid for stock repurchases and that did not decreased slightly before the crisis, then rose by 1% during
2008-36 2010 and then decreased again (by 1.3%). In all the periods it was primarily because of the the changes in companies characteristics. It confirms the hypothesis that companies with stock repurchases held more cash during and after the crisis than the rest of the sample.
Table 8. Components of change in cash ratio difference
This table shows the contribution of observed characteristics (2), coefficients (3) and the unobservables (4) to the increase in gap of cash savings between groups. Total change is the change in gap between 2 groups over time.
Differential Total Change (1) Observed Quantities (2) Observed Prices (3) Unobserved Prices and Quantities (4)
Panel A. Domestic & Multinationals
2000-2008 -0.022 -0.019 -0.031 0.028 2000-2010 -0.024 0.000 -0.022 -0.002 2000-2013 -0.044 0.025 -0.032 -0.037 2008-2010 -0.002 0.015 0.012 -0.029 2008-2013 -0.022 -0.020 0.063 -0.065 2010-2013 -0.02 0.012 0.004 -0.035
Panel B. Constrained & Unconstrained
2000-2008 0.006 0.151 -0.015 -0.131 2000-2010 0.008 0.073 -0.039 -0.025 2000-2013 0.012 -0.046 -0.116 0.174 2008-2010 0.002 -0.104 0.001 0.105 2008-2013 0.006 -0.199 -0.1 0.304 2010-2013 0.004 -0.098 -0.097 0.199
Panel C. Dividend payers & Dividend non-payers
2000-2008 0.026 -0.018 0.022 0.022 2000-2010 0.035 -0.016 0.011 0.04 2000-2013 0.029 -0.009 0 0.038 2008-2010 0.01 0.012 -0.02 0.018 2008-2013 0.003 -0.008 -0.005 0.016 2010-2013 -0.007 -0.002 -0.003 -0.002
Panel D. Stock Repurchases & No stock repurchases
2000-2008 -0.006 -0.023 0.013 0.004 2000-2010 0.003 -0.018 0.006 0.015 2000-2013 -0.01 -0.03 0.002 0.017 2008-2010 0.009 0.019 -0.021 0.011 2008-2013 -0.004 -0.023 0.005 0.013 2010-2013 -0.013 -0.015 0 0.002
37 6. Robustness Checks
Robustness checks are intended to examine how certain coefficient estimates behave when the regression specification is modified by changing, removing or adding explanatory variables. If the coefficients are robust, it is usually interpreted as a structural validity. Robustness checks may also involve changes in empirical specifications, sampling restrictions, and econometric methodologies.
First, the regression for the first hypothesis is checked, where several variables are calculated differently than in the initial model. Table 9 shows these modifications. The first column duplicates the third column from Table 4, based on which the predicted cash holdings were estimated and then compared to the actuals. Column (2) is almost the same as the first model, except for the dependent variable is calculated as a natural logarithm of cash over assets. It is clear from the table that two core variables become insignificant (Cash Flow and Capex). However, the direction of the relationships between cash ratio and significant explanatory variables is the same for both models. Column (3) represents the regression results where tangible assets replace total assets when calculating several variables: cash ratio, net working capital, leverage and capital expenditures. This specification gives a higher Adjusted R2 (36% against 31% in column (2)), whereas such coefficients as market-to-book ratio and size become insignificantly different from zero. The rest variables influence cash holdings in the same direction as in the models (1) and (2).
Thus, I can conclude that the model which estimators were used to predict cash holding during and after the financial crisis is robust to changes in model specifications.
The second robustness test is for the second bunch of hypotheses. Three panel regressions were built for each of the periods that will be compared to those obtained in Table 7. It is seen in Table 10 that before the crisis multinationals had on average 1.8% higher cash ratio than the rest of the sample, firms with stock repurchases held 1% more cash that firms that did not. On the other hand, dividend-payers as well as unconstrained companies kept less money than their counterparts. To sum up, the hypotheses are proved on the pre-crisis sample.
38 Column (2) reflects the crisis period that includes 2008 and 2009 years. As in Table 7, market-to-book ratio and size are not significant in this period of time, as well as multinationals and companies with stock repurchases. These results and the signs of direction for other explanatory variables are identical to those obtained in Table 7. Column (3) presents results for the post-crisis period. Most of the coefficients have the same sign of direction with a dependent variable, whereas multinational dummy that was not significant in any distinct year of this period, became significant at 1% level. It means that multinationals have 1.1% higher cash ratios than purely domestic firms after the 2008 financial crisis.
Table 9. Robustness checks for the base period
This table displays the regression models for the base period (2000-2006). Each model includes industry fixed effects, year fixed effects and clustered standard errors. In column (3) such variables as net working capital, capital expenditures and leverage are calculated using tangible assets in the denominator. Column (1) fully repeats the results obtained in column (3) of table 4. The dependent variable in column (2) is calculated as logarithm of cash over total assets and in column (3) as cash divided by tangible assets. *, **, *** indicate that the coefficient is significantly different from zero at the 10, 5 and 1% level, respectively.
Model (1) 2((( (2) (3)
Dependent variable Cash/Assets Ln(Cash/Assets) Cash/Tangible Assets
Q 0.001*** (0.000) 0.009*** (0.003) -7.9E-06 (0.000) Size -0.006*** (0.001) -0.050*** (0.009) 0.001 (0.001) Cash Flow -0.016*** (0.001) 0.001 (0.011) -0.015*** (0.001) NWC4 -0.021*** (0.004) -0.267*** (0.033) 0.012*** (0.003) Capex5 -0.011*** (0.002) -0.021 (0.021) -0.029*** (0.003) Leverage6 -0.070*** (0.004) -0.633*** (0.033) 0.000 (0.000) R&D 0.037*** (0.002) 0.199*** (0.019) 0.049*** (0.003) M&A -0.014*** (0.001) -0.089*** (0.013) -0.010*** (0.002) Div dummy -0.013*** (0.004) -0.077*** (0.029) -0.020*** (0.004) R&D dummy 0.018*** (0.006) 0.184*** (0.044) 0.020*** (0.006) M&A dummy -0.016*** (0.002) -0.098*** (0.021) -0.015*** (0.003) Constant 0.206*** (0.056) -2.534*** (0.388) 0.155** (0.061) Adj R2 0.37 0.31 0.36
4 In Model 3, the variable is calculated as Net Working Capital to tangible assets
5 In Model 3, the variable is calculated as Capital Expenditures to tangible assets
6
39 Table 10. Robustness checks for sub-samples.
The table displays panel regressions for the pre-crisis period (2000-2006), crisis (2008-2009) and post-crisis period (2010-2013). Each model includes industry fixed effects, year fixed effects and clustered standard errors. *, **, *** indicate that the coefficient is significantly different from zero at the 10, 5 and 1% level, respectively.
Variable Pre-crisis period
(1) Crisis period (2) Post-crisis period (3) Market-to-book ratio 0.001** (0.000) -0.001 (0.001) 7.0E-05 (0.000) Size 0.001 (0.002) 0.000 (0.002) -0.003** (0.002) Cash Flow -0.016*** (0.001) -0.012*** (0.002) -0.010*** (0.001) NWC -0.020*** (0.004) -0.021*** (0.002) -0.001 (0.001) Capex -0.011*** (0.002) -0.027*** (0.003) -0.012*** (0.002) Leverage -0.069*** (0.004) -0.123*** (0.008) -0.030*** (0.003) R&D 0.037*** (0.002) 0.040*** (0.003) 0.039*** (0.002) M&A -0.015*** (0.001) -0.011*** (0.002) -0.016*** (0.001) R&D dummy 0.021*** (0.006) 0.018** (0.008) 0.019*** (0.006) M&A dummy -0.017*** (0.002) -0.017*** (0.004) -0.011*** (0.003) Volatility 0.004*** (0.000) 0.003*** (0.000) 0.003*** (0.000) Multinational dummy 0.018*** (0.003) 0.008 (0.006) 0.011*** (0.004) Repurchase dummy 0.008*** (0.002) -0.002 (0.003) 0.005** (0.002) Dividend dummy -0.012*** (0.004) -0.018*** (0.005) -0.012*** (0.004) Unconstrained dummy -0.036*** (0.005) -0.032*** (0.009) -0.027*** (0.005) Constrained dummy -0.002 (0.005) 0.011 (0.008) -0.002 (0.006) Constant 0.149*** (0.056) 0.163*** (0.05) 0.260*** (0.044) Adj R2 0.38 0.42 0.41
40 7. Conclusion
This thesis examines the determinants of corporate cash holding among public U.S. companies from 2000 to 2013. It was investigated that during the financial crisis and right after it American firms held abnormally high levels of cash. In this context abnormal cash holdings are cash savings that cannot be explained by patterns from 2000-2006 years (pre-crisis period). Namely, the abnormal cash-to-assets ratios were 5.4% and 6.4% in 2008-2009 in 2010-2013 respectively. Moreover, it was found that such a growth in cash is due to companies that do not pay dividends, repurchase stock and are financially constrained. There is no evidence that multinational companies hold more cash abroad because of the tax incentives. Furthermore, using Juhn-Murphy-Pierce approach it was proven that the difference in cash savings between international and purely domestic firms has been decreasing over time especially from 2008 to 2013.
Moreover, the findings regarding the main cash ratio determinants are consistent with the existing literature. Market-to-book ratio, R&D expenses and industry volatility have a positive influence on cash savings. Furthermore, cash ratio significantly decreases with firm size, cash flow, net working capital, leverage, capital expenditures and M&A.
The results of this thesis in relation to the stated hypotheses are partly in line with the related literature. First of all, the result that cash savings were abnormally high after the financial crisis is in agreement with the results obtained in the paper by Pinkowitz et.al. (2013). However, the authors attribute this increase in cash to multinational companies and most profitable firms, while I found that cash savings of multinationals did not significantly differ from the rest of the sample.
According to Almeida, Campello and Weisbach (2004), financially constrained companies hold more cash as they have limit access to the capital markets. In contrast, it was investigated in the thesis that financially unconstrained companies keep less cash on their balance sheets.
