The impact of R&D on cash holdings in
multinationals – Evidence from the UK
Master Thesis
Jonas Becker
University of Groningen: S2339889
Uppsala University: 880525-P439
Email: jonas.beckerbs@googlemail.com
Supervisor: Dr. J. H. von Eije
Assessor: Dr. W. Westerman
1 Abstract
This paper1 investigates whether R&D influences cash holdings differently for multinationals than for domestic companies. Theory suggests that R&D impacts cash holdings positively in general and might do so to a bigger extent for multinationals, mainly because firms can better protect their property rights on intangible assets abroad by exploiting them directly themselves and cash holdings can assist in that. Although confirmed by Pinkowitz, Stulz and Williamson for their US sample looking at the periods 1998-2000, 2004-2006 and 2009-2010, my UK sample of 2000-2012 taken from Datastream does not yield a significant interaction term. The results point to yet unexplored country specific reasons within the UK, or the relationship being a US-specific phenomenon.
Key words: cash holdings, research and development, multinationality, United Kingdom
JEL classification: F23; G32; O32
1
2 1. Introduction
During the last three decades companies’ cash holdings have more than doubled on average (Bates, Kahle and Stulz, 2009). This might seem counterintuitive, as an increase in financial and information technology should reduce transaction costs, agency conflicts and costs of obtaining external financing (Ross, 1989). Moreover, certain macroeconomic forces, namely the liberalization of financial markets and lower currency volatility, should reduce transaction and bankruptcy costs for multinationals through centralized cash management activities (Westerman and Von Eije, 2005), both of which would suggest a decrease in cash holdings. A major reason may be, that idiosyncratic risk has been rising historically, which leads to firms raising cash levels (Bates, Kahle and Stulz, 2009).
Recently, Lyandres and Palazzo (2012) explored that the mean cash-to-assets ratio of firms belonging to the top quintile of R&D-to-assets ratio has increased from about 10% to about 50% between 1976 and 2010 while the ratio for the bottom quintile only increased from about 12% to about 17%. At the same time, the mean R&D-to-assets ratio in the top quintile has increased from 10% to 45% while remaining constant at zero in the bottom quintile. Similarly, Zhou (2009) found that from 1980 to 2007 cash holdings in high-tech firms increased from 11% to 39% of total assets, while cash holdings in non-high-tech firms remained stable at around 11%. He argues that this is due to many new firms going public in this timeframe, because of more developed capital markets new firms could have weaker fundamentals and this led to a rapidly expanding tech sector. This transformed the high-tech sector to having more characteristics closely related to a need for high cash holdings (Zhou, 2009).
3
holdings is found. Additionally, firms appear to not hold more cash after becoming multinationals, but rather before they do.
This strongly points to a connection of R&D and multinationality, however Pinkowitz, Stulz and Williamson (2012) do not report numbers on the stronger relationship between R&D and cash holdings in US multinationals compared to US domestic firms and suggest to do further work on this issue. This paper investigates the reasons why multinationals hold more cash than domestic companies by aiming to establish the degree of multinationality as a mediator for the impact of R&D on cash holdings. Pinkowitz, Stulz and Williamson’s US focus is especially of interest, as the surge in cash holdings from the 1990s till the crisis is US made, meaning that foreign firms do not experience a comparable increase, while the rise from pre- to post-crisis continues throughout their sample (Pinkowitz, Stulz and Williamson, 2013). To avoid a US bias, I investigate a UK sample, which Pinkowitz, Stulz and Williamson (2012) found to be unique enough to warrant an analysis separated from other non-US advanced economies, because of its similarity to the US. A number of important country characteristics likely influencing the link between R&D, multinationality and cash holdings namely shareholder protection, country risk, corruption and cultural factors such as uncertainty avoidance are quite comparable between the UK and the US. In combination with the UK’s exports to Europe this leads to the UK being a prime candidate to test whether the stronger relationship not only holds in the US.
4
large amount of cash holdings. This reasoning leads to a stronger relationship between R&D and cash holdings in multinationals than in domestic firms.
The intuition of the argument is tested by conducting panel regressions of UK firms over the years 2000-2012. First, cash holdings are regressed on the firms’ R&D expenditures scaled by total assets, their multinationality measured as foreign sales to total sales and the interaction of R&D and multinationality, using fixed effects. Second, the same test is performed using random effects to deal with problems possibly arising from the interdependence of observations.
The findings do lead to a rejection of the hypothesis that there is a stronger relationship between R&D expenses and cash holdings in multinationals than in domestic companies. The fixed effects panel regression does not yield significant results for either the interaction term, or the individual variables R&D and multinationality. Similarly, the random effects regression does not show a significant interaction term of R&D and multinationality. It does however uncover a major characteristic of the variables, namely that cross-section effects dominate time-series effects within a firm. This is shown by conducting a Fama-MacBeth regression (Fama and Fama-MacBeth, 1973), which gives significant and positive results for the main variables individually. The results regarding the interaction term are robust when testing for non-linearity, the behavior as multinational compared to domestic firms and high-tech compared to non-high-high-tech firms as opposed to the degree of multinationality and R&D intensity using dummies, the impact of R&D on cash holdings of firms prior to becoming multinationals, multinationality-quantiles and multicollinearity. Both the robustness test testing the impact of R&D on cash holdings of firms prior to becoming multinationals, which gives a significant and positive multinationality term and the non-linearity test point to firms with an already high level of international sales increasing their cash holdings more with further internationalization of sales and, given a lower initial level, less so.
5
future studies of cash holdings and their determinants. As for managerial implications, decision-makers would have benefited from positive findings in that it would have helped create awareness of the increased need and value of cash holdings for domestic firms on the verge of becoming multinationals, especially as domestic companies in the US do not increase their cash holdings at the point of becoming multinational companies, but before this process (Pinkowitz, Stulz and Williamson, 2012). Considering the non-existent link in the UK as opposed to the US despite their similarities though, the relationship cannot be generalized and appears to be either a US-specific phenomenon, or there are yet unexplored country specific reasons within the UK. The latter explanation is supported by the recent trends in the UK, with mean cash holdings stagnating and mean R&D falling.
The paper will be structured as follows. After this introduction, in chapter 2 the existing theoretical background regarding cash holdings and the links between them and R&D and multinationality is presented and the hypothesis is developed. Chapter 3 displays the data and methodology used to test the theoretical considerations. Subsequently, chapter 4 presents the results and robustness tests. Finally, chapter 5 concludes and gives limitations and suggestions for further potential directions of research.
2. Theoretical Background and Hypothesis
6
Building on the first two theories, Opler, Pinkowitz, Stulz and Williamson (1999) studied the marginal benefits and costs of cash and found that certain firm characteristics explain the level of cash holdings in firms. These include cash flow volatility, growth opportunities, capital expenditures, cash flows, leverage, non-cash forms of liquidity and access to them, firm age and firm size. In the next sections, a closer look is taken at the firm characteristic R&D intensity, the degree of firms’ multinationality and their interaction effects, and how these can affect cash holdings.
2.1 Cash holdings and R&D
Opler, Pinkowitz, Stulz and Williamson (1999) also expand the transaction cost motive by positing the tradeoff theory. This theory considers transaction costs, but adds the concepts of asymmetric information and agency costs for external financing. While I will consider the latter in the next subsection regarding cash holdings and multinationality, the former is relevant for R&D expenses. Reasons why Modigliani and Miller’s theorem (1958, 1961) does not hold in practice include that R&D investments are inherently uncertain, which in combination with asymmetric information favors a real options approach and that cost of capital might differ for different sources for tax, non-tax and type of investment reasons (Hall, 2002).
Regarding the increased uncertainty, Suurmeijer, Smid and Von Eije (2013) found not only systematic cash flow risk, but also unsystematic or idiosyncratic risk, beta and total risk increasing with more R&D investments. Spending on R&D increases risk and risk increases the need for cash holdings (Bates, Kahle and Stulz 2009). This points further to a positive relationship between R&D and the level of cash holdings.
As for type of investment reasons, in contrast to investment in capital, R&D investment creates mainly intangible assets, with little to no collateral value (Himmelberg and Petersen, 1994). Recent empirical findings confirm that firms in general, but especially financially constrained ones, make larger adjustments to cash holdings when they plan additional future R&D rather than fixed capital investment expenditures (Baum, Caglayan and Talavera, 2012).
7
counteract this, firms use more cash holdings to smooth R&D investment in bear markets and use bull markets to build cash holdings (Shin and Kim, 2011a). Holding cash is expensive in itself, but less so than utilizing external funds (Brown and Petersen, 2011). In order to safeguard against negative shocks to the internally generated cash flows and the external equity financing availability, these firms therefore have to hold sufficient cash holdings if they do not want to risk having to cut their R&D spending.
Adjusting the flow of R&D is very expensive, as Brown and Petersen (2011) comprehensively illustrate. A temporary cut to R&D spending would likely mean releasing employees, because most of the investment in this area is tied to wages. These workers would have to be rehired when the spending resumes at its former level. However, next to the hiring costs, they typically require firm-specific training, often work in teams which should not be disrupted and are in possession of highly sensitive proprietary information. As Baum, Caglayan and Talavera (2012) note, R&D capital stock is represented by human capital, assets are often tacit as opposed to codified, thus temporarily reducing expenditures would lead to this human capital being lost to other companies, in turn leading to the competitors gaining the sensitive information. Therefore, adjusting R&D is likely much more costly than adjusting physical investment and firms are enticed to smooth their investment. As an example, while a firm could reschedule a multi-year purchase of machinery, reducing R&D temporarily would be much more difficult (Hall and Lerner, 2009). Consequently, cash holdings can give firms better control over present and future adjustment costs of an ongoing R&D program (Brown and Petersen, 2011).
That being said, recent research points to higher innovation efficiency given financial constraints. Financial constraints can mitigate free cash flow problems, especially in firms with high excess cash holdings and low investment opportunities. Consequently, these firms then refrain from investments in fields out of their direct expertise (Almeida, Hsu and Li, 2013). So while financial constraints reduce R&D investment, the remaining investment is of higher efficiency. Lyandres and Palazzo (2012) not only find a direct strategic reason for high R&D firms to have high cash holdings to deter other firms from investing, but also associate this innovation efficiency with increased cash holdings because of these strategic considerations.
8
investments, increasing bankruptcy costs, mainly to the detriment of debt-holders. Also, proprietary information needs to be protected, even from potential investors (Brown, Fazzari and Petersen, 2009), which further accentuates the moral hazard problem. Obviously, this increases cost of debt and is an argument for equity financing and against debt for R&D investments. This is to some extent mitigated by government interference (subsidies, incentives, intellectual property rights) and venture capital, but still external funding is requiring an abnormally high return on capital because of asymmetric information, moral hazard and tax considerations (Hall, 2002). Thus, in order to increase cash holdings, firms increasingly issue shares. While in the 1970s 1$ of issuance resulted in 0.23$ of cash, in the 2000s 0.60$ of the 1$ issued went into cash holdings (McLean, 2011). Financing in general is especially important for young, R&D-intensive firms, as cash flow and stock issues are highly volatile in this type of firm, while debt financing is prohibitively expensive due to information problems, uncertain returns and lack of collateral (Brown, Fazzari and Petersen, 2009). Equity clearly seems to be the primary source of funds for R&D-intensive firms. However, despite its advantages regarding R&D financing, Brown and Petersen (2011) also indicate shortfalls, namely high flotation costs of public stock issues and a “lemons premium”, stemming from asymmetric information. Furthermore, going public to gain access to lower-cost equity capital is similar to debt negatively influenced by the impact of losing confidential information to competitors (De Jong, Huijgen, Marra and Roosenboom, 2012).
In a recent paper, Keefe and Kieschnick find the market value of a 1$ increase in cash holdings to be 0.256$ higher for small firms that engage in R&D, compared to small firms that do not (Keefe and Kieschnick, 2011). Relatedly, young firms use more cash holdings to smooth R&D investment than mature ones (Shin and Kim, 2011b). For small and new innovative firms cost of capital for financing R&D is high, for large firms less so, but they still prefer internal funding as opposed to external funding and manage their cash flows to ensure this (Hall and Lerner, 2009). This is in line with the findings of Himmelberg and Petersen (1994), who identified the flow of internal finance to be the main determinant of the rate at which small, high-tech firms acquire technology through R&D. Consequently, certain innovations, which would have been provided with private capital, will not be accessible at the external cost of capital hurdle if the firm is unable to raise that capital internally (Hall, 2002). Finally, firms with higher cash holdings are more likely to win patent races than firms with less cash (Schroth and Szalay, 2010).
9
Opler, Pinkowitz, Stulz and Williamson (1999) consider agency costs and their effects on cash holdings within their tradeoff theory. The agency theory ascertains that high free cash flow leads to agency problems (Jensen, 1986). In the context of cash holdings, this leads to managers hoarding cash rather than paying dividends. Therefore, with weak external shareholder protection high cash holdings lower firm value and paying dividends increases firm value (Pinkowitz, Stulz and Williamson, 2006 and Kalcheva and Lins, 2007). Still, weak external shareholder protection leads to firms in these countries to hold significantly larger amounts of cash because of managerial preferences (Dittmar, Mahrt-Smith, and Servaes 2003). Similarly, internal governance is an important factor determining the value of cash holdings. 1$ of cash in a poorly governed firm has only about half the value than cash in a well-governed firm (Dittmar and Mahrt-Smith, 2007). The suggestion that firms hold more cash in countries with greater agency problems likely has an influence on multinationals, as for firms from countries with comparably high shareholder protection and good internal governance mechanisms going abroad with subsidiaries decreases these attributes on average and vice versa. Next to shareholder protection, a similar effect on the average values of country risk and corruption can be expected (Dittmar, Mahrt-Smith, and Servaes, 2003), again influencing cash holding levels. Additionally, it is also important to consider the reported effect of culture on cash holdings. Certain country-specific cultural factors can influence cash holdings, with e.g. high uncertainty avoidance country scores leading to firms in these countries to hold more cash on average. Multinationality moderates this effect in a similar way as previously described (Ramirez and Tadesse, 2009).
Ramirez and Tadesse (2009) argue that an extended business cycle, i.e. goods taking longer to be produced and shipped to the point of sale, longer decision-making processes, etc., a complex network of subsidiaries and the tax motive lead to multinational firms holding more cash than domestic firms and they give empirical support for this. The latter is based on Foley, Hartzell, Titman and Twite’s suggestion (2007) that multinational corporations might hoard cash due to tax costs of profit repatriations. However, Pinkowitz, Stulz and Williamson (2012) do not find evidence that the increase in cash holdings of multinational firms could be explained by the tax motive.
10
multinationals negatively, as cash management activities become increasingly centralized and transaction and bankruptcy costs for multinationals have been reduced. Furthermore, higher corporate diversification is associated with less cash holdings, as it can serve as a way to economize on cash, because outcomes and opportunities of the divisions are not perfectly correlated, thus reducing risk exposure (Duchin, 2010). Diversified firms can then hold a lower aggregate, joint amount of cash at the headquarters level. This does hold even after controlling for economies of scale (Mulligan, 1997).
Considering this paper’s focus on the UK, I would expect a positive relationship between multinationality and cash holdings in the UK, given the comparably high shareholder protection, good internal governance, low country risk, low corruption and low uncertainty avoidance levels of firms in the UK. Its similarity to the US in these characteristics is expected to outweigh the opposing forces.
2.3 R&D and Multinationality
The main existing theoretical link between R&D and multinationality is that with R&D being a major source of intangible assets, firms look to better protect their property rights on these abroad by exploiting them directly (Morck and Yeung, 1991). Since this is easier for multinational corporations than for domestic companies, Morck and Yeung (1991) found it to be valued more highly. Furthermore, they find no isolated impact of multinationality on firm value, but they do find an increased positive impact of R&D on firm value given an increase in multinationality; the more multinational a firm, the more positive impact does spending on R&D have on firm value. Obviously, this is in line with the finding that multinationals spend more on R&D.
11
growing, be it on the product or the market level, they can use their resources more efficiently to maximize gains from competitive advantages gained through R&D. A third reason for linking R&D and multinationality is, that firms with R&D capabilities are better able to realize the inherent benefits of multinationality, since they can charge premiums for their technologically differentiated products in more markets or they can apply their improved manufacturing processes with lower production costs or better product quality to achieve economies of scale (Kotabe, Srinivasan and Aulakh, 2002).
2.4 Cash holdings, R&D and Multinationality
Evidence from Pinkowitz, Stulz and Williamson (2012) shows that the firms which become multinationals after 1998 have high cash holdings when they do so. These results suggest that multinational firms and those looking to become one have unique attributes that make cash holdings particularly valuable for them. The theoretical considerations and empirical evidence presented in the last sections point to R&D at least being one of these attributes, as access to R&D needs to be protected by having a relatively large amount of cash holdings. In their sample, R&D-intensive domestic firms increased their cash holdings compared to non-R&D domestic firms and similarly no R&D multinational corporations increased their cash holdings compared to non-R&D domestic firms. However, both of these increases are quite small, especially compared to the increase of cash holdings for R&D-intensive multinationals, be it compared to R&D-R&D-intensive domestic firms or to non-R&D multinationals. The authors further regress abnormal cash holdings on the interaction term of R&D and multinationality and the two individual variables, finding a positive significant effect of the interaction term. In an effort to replicate these results for a UK sample, the following set of hypotheses is developed:
H0: There is no different relationship between R&D expenses and cash holdings in
multinationals than in domestic companies.
H1: There is a stronger relationship between R&D expenses and cash holdings in
multinationals than in domestic companies.
12
If H1 cannot be rejected, this would help explain the rather recent enormous rise in cash holdings. With multinationality steadily increasing, the R&D expenditures would have an increasingly big, positive impact on the level of cash holdings. Consequently, managers could make better informed decisions on the optimal level of cash holdings prior to internationalizing their company and academics would have to account for the interaction effect in empirical studies regarding cash holdings. The next chapter illustrates how this relationship is tested.
3. Data and Methodology 3.1 Data
This paper is investigating a sample of listed UK-based firms. I use data from the Datastream database. The selection process includes active, as well as dead and suspended listings in order to avoid survivor bias. Companies with usable International Securities Identifying Number (ISIN) and industry codes (SIC) are selected, while companies with similar ISIN codes and similar names, companies that give error codes in downloading data, and companies that report in currencies other than British pounds are eliminated. Because utility (SIC codes 4900-4999) and financial (SIC codes 6000-6999) firms might be subject to regulatory supervision or might carry cash to meet capital requirements rather than for the reasons investigated here, these firms are excluded and the primary focus in this paper is on industrial and service firms. After adjustments, this procedure informs on cash holdings for 3021 firms, yielding 19791 firm-year observations for the period 1999-2012. Furthermore, as a measure for scale, all financial data are divided by total assets. In order to avoid endogeniety, the independent variables are lagged by one year. Finally, to reduce the effect of outliers, extreme values are winsorized at the 1st and 99th percentiles.
3.2 Main and Control Variables
13
Fig.1 The development of the mean of cash holdings (CH), research and development expenditures (RD) and
multinationality (MN), each scaled by total assets, UK, 1999-2012. Cash holdings are cash and equivalents, R&D is R&D expenses and multinationality is foreign sales divided by total sales. The variables are winsorized at the 1st and 99th percentiles.
Above, figure 1 displays the development of the mean of the three main variables cash holdings (CH), research and development (RD) and multinationality (MN) from 1999 to 2012 in the UK. Mean cash holdings were 14.8% of total assets in 1999, peaked at 21.9% in 2005 and subsequently decreased to 18.2% in 2012. Mean R&D expenditures were 7.7% in 1999, almost doubled until 2003 reaching 14%, but then decreased until the end of the sample period, hitting the low point of the sample with 5.1% in 2012. Mean multinationality however, steadily increased from 25.5% in 1999 to 48.8% in 2012, almost doubling in 13 years.
Research identifies various determinants and motives for cash holdings that have to be accounted for when investigating the impact of R&D intensity and multinationality. This paper closely follows Bates, Kahle and Stulz’ model (2009) for obtaining estimates of cash holdings. In the following paragraphs, the control variables are presented and their expected impact is described. A comprehensive table of all the variables used can be found in Appendix A.
First, Opler, Pinkowitz, Stulz and Williamson (1999) argue that risk (RI) is one of the main factors influencing cash holdings positively, since uncertainty leads to situations where the firm has more outlays than expected at times. In such a situation, firms with greater cash flow uncertainty should be holding more liquid assets and thus also more cash. The precautionary theory of cash holdings arrives at the same conclusion (Keynes, 1937). As a
14
measure for risk, the standard deviation of the operating income divided by the total assets for the last two years and the current year is used.
Furthermore, growth opportunities (GO) impact cash holdings positively as possible underinvestment problems (Myers, 1977) are mitigated by higher cash holdings. For these firms adverse shocks and financial distress are more costly and therefore the precautionary motive dictates them to hold more cash (Opler, Pinkowitz, Stulz and Williamson, 1999 and Keynes, 1937). In line with Fama and French (2001), these growth opportunities are measured by the sum of total assets and the market value of equity minus the book value of equity, over total assets.
Similarly, cash flows (CF) are assumed to influence cash holdings positively, as financially constrained firms invest in cash out of cash flows, and cash from high cash flow states of the world is being transferred to fund investment in all states, including low cash flow ones (Bates, Kahle and Stulz, 2009). Also, as cash flows are a determinant of investment spending (Fazzari, Hubbard and Petersen, 1988), higher cash flows lead to higher planned expenditures in capital and R&D, and therefore higher cash holdings are needed. That being said, Pinkowitz, Stulz and Williamson (2012) find a significant negative influence of cash flows on cash holdings. Intuitively, a reason could be that firms with high cash flows feel less need to have high cash holdings, as they are able to fund investments out of cash flows without having to resort to cash holdings. As a proxy for cash flows, operating income divided by total assets is used, because a comparable number of observations on operating cash flows is not available in the Datastream database.
Although capital expenditures (CE) were found to be comparably easy to finance and to generate assets that can be used as collateral, therefore having less of an impact than R&D expenditure (Bates et al. 2009), they still do consume cash and thus are considered as a control variable. CE are measured as capital expenditures divided by total assets.
15
level of cash due to managerial preferences (Jensen, 1986). The variable LE consists of total debt value scaled by total assets.
Obviously, liquidity (LQ), which is going to be measured by receivables and inventory, is a substitute for cash and thus influencing its level negatively (Bates, Kahle and Stulz, 2009). Originally proposed by Keynes (1936), the big advantage of liquidity, namely being able to undertake valuable projects whenever they arise, is less of importance when firms are financially unconstrained and have access to external capital. Consequently, short-term access to liquidity is considered, as firms may have access to short short-term bank loans, which would reduce the need for cash holdings (Von Eije, 2012 and Almeida, Campello and Weisbach, 2004). Short-term debt divided by total assets represents this variable.
Furthermore, firm size (FS) serves as a proxy for access to capital markets (Opler Pinkowitz, Stulz and Williamson, 1999), and it is an important factor because of economies of scale, which both effectively reduces the need for cash holdings (Mulligan, 1997). It is measured using the natural logarithm of total assets. Since young firms’ target cash ratio is determined largely by the precautionary savings motive and older firms adjust cash ratios much slower and do not regard the precautionary savings motive as much (Dittmar and Duchin, 2010), firm age (FA) measured by the difference between the current year and the date of incorporation is also considered.
16 3.3 Regressions
The following fixed effects and, to deal with problems possibly arising from the interdependence of observations, random effects panel data regressions with cash holdings (CH) as the dependent variable are conducted:
, = + ( , ∗ , ) + , + , + , + ,
+ , + , + ! , + "!#, + $!, + $,
+ %, + , + &,
Where
, = dependent variable, cash holdings
, ∗ , = interaction variable, research and development and multinationality
, = independent variable, multinationality
, = independent variable, research and development
, = control variable, risk
, = control variable, growth opportunities
, = control variable, cash flow
, = control variable, capital expenditures
! , = control variable, leverage
!#, = control variable, liquidity
$!, = control variable, short-term access to liquidity
$, = control variable, firm size
% , = control variable, firm age
, = control variable, dividends
&, = error term.
17 3.4 Summary Statistics
Below, table 1 displays the descriptive statistics of all variables used.
Table 1 Sample Descriptive Statistics. CH is cash holdings. MN is multinationality. RD is research and
development. RI is risk. GO is growth opportunities. CF is cash flow. CE is capital expenditure. LE is leverage. LQ is liquidity. SL is short-term access to liquidity. FS is firm size. FA is firm age. DI is dividends. All financial data are calculated as ratios to the firm’s total assets. All variables except for firm size and firm age are winsorized at the 1st and 99th percentiles. All variables except for the dependent variable cash holdings and firm age are lagged by one year.
Mean Median Maximum Minimum Std. Dev. Observations
CH 0.192 0.1 0.97 0 0.232 19791 MN 0.351 0.227 1 0 0.359 14126 RD 0.095 0.026 1.086 0 0.177 6899 RI 0.127 0.035 2.435 0.002 0.321 16811 GO 2.366 1.424 24.19 0.48 3.211 18101 CF -0.086 0.041 0.372 -2.926 0.448 19745 CE 0.049 0.026 0.397 0 0.068 19424 LE 0.193 0.122 1.603 0 0.256 19774 LQ 0.283 0.244 0.919 0 0.224 19529 SL 0.073 0.022 0.877 0 0.136 19427 FS 10.473 10.344 16.682 4.6 2.4 19801 FA 29.635 17 156 0 31.453 33719 DI 0.015 0 0.146 0 0.025 19448
18 4. Results
4.1 Regression Results
Both the fixed effects and the random effects regression yield results that lead to a rejection of H1. Table 2 displays those findings. The interaction term has a negative coefficient and is insignificant, indicating no different relationship between R&D and cash holdings in multinationals compared to domestic firms. Interestingly, both multinationality and R&D are individually insignificant in the fixed effects regression. This means that an increase in R&D expenditure or multinationality does not have significant time-series effects on cash holdings. R&D is significant and positive in the random effects regression, though. However, the Hausmann Test, the results of which are not displayed here, is significant at the 1% level, suggesting that fixed effects are to be preferred. In order to test whether cross-sectional effects might dominate panel effects for R&D as an explanatory variable for cash holdings, a Fama-MacBeth regression is performed as a robustness test in the next section.
In the fixed effects regression results, the control variables risk, growth opportunities, capital expenditure, liquidity and firm size are significant at the 1% level, while, leverage and short-term access to liquidity are significant at the 5% level. Furthermore, all of the coefficients have the expected direction of impact. However, cash flow, dividends and firm age do not have a significant effect on cash holdings. Considering that cash flows have been found by Pinkowitz, Stulz and Williamson (2012) to have the opposite direction of previously existing empirical evidence, this result is not too surprising. Dividends stay insignificant when separately looking at non-dividend and dividend payers (the results of the additional test are not reported here). Firm age is insignificant because of the inclusion of year dummies. Omitting these, firm age is significant at the 5% level and the coefficient negative, in line with theory. As for the random effect regression, other than the difference regarding the R&D variable, all variables have the same direction of impact and the significance of leverage increases to the 1% level.
19
Table 2 Results for the panel regressions with fixed effects and random effects for the full sample of firms from
the UK for 2000-2012 . The dependent variable is cash holdings. C indicates the coefficient of the variable and P the level of significance. * indicates a significance level of 10%, ** of 5% and *** of 1%. RD*MN is the interaction variable of research and development and multinationality. MN is multinationality. RD is research and development. RI is risk. GO is growth opportunities. CF is cash flow. CE is capital expenditure. LE is leverage. LQ is liquidity. SL is short-term access to liquidity. FS is firm size. FA is firm age. DI is dividends. C is the coefficient. Observ. is the total amount of observations for which all information was available. Adj. R² is the adjusted R-squared. Prob. indicates the significance of the equation. All financial data are calculated as ratios to the firm’s total assets. All variables except for firm size and firm age are winsorized at the 1st and 99th percentiles. All variables except for the dependent variable cash holdings and firm age are lagged by one year. Time dummies for the years 2002-2012 have been added, but are not shown.
Fixed Effects Random Effects
20
Nevertheless, the mean cash holdings-to-assets ratio for the UK firms in the top multinationality quartile is 18.17% compared to 18.15% in the bottom multinationality quartile for the whole sample, which points to an irrelevance of multinationality. It is also important to consider that the UK moved from taxing income worldwide to only taxing income domestically in 2009, which should lead to multinational firms raising cash holdings post 2009 in their subsidiaries in order to save repatriation tax expenses (Foley, Hartzell, Titman and Twite, 2007). In line with this, the mean cash holdings of UK multinationals increase from 15.9% in 2009 to 17.3% in 2010, however subsequently decline again to 15.5% in 2012. Regarding R&D expenditure, the mean dropped from 14% in 2003 to 5.1% in 2012 and the median actually fell from 4.5% to 0% in the same time period. This development is quite surprising, especially considering R&D should be valued more highly in multinationals than in domestic firms, according to Morck and Yeung (1991).
One possible explanation for the different results compared to the US is yet unexplored country specific reasons. Taking into account the aforementioned trends, the UK does seem to differ from the rest of the world and the US. The exact reasons for the differences require further research. Another possible explanation is that the stronger relationship between R&D expenses and cash holdings in multinationals than in domestic companies is a US-specific phenomenon. Pinkowitz, Stulz and Williamson (2012) note an increase in cash holdings from the 1990s to the pre-crisis and a total level of cash holdings that is unique to the US, while the development of cash holdings from before the crisis to after the crisis is not. On request, Mr. Pinkowitz supplied the results from their regression with the interaction term. However, he noted that there is no data available confirming the interaction of R&D and multinationality and its effects on cash holdings for other countries than the US. Therefore, this relationship could still be US-specific, even while the changes in cash holding levels have been normalized.
4.2 Robustness Tests
In order to examine the validity of my results, several robustness tests are performed. The tests are checking the robustness of the fixed effects regression and not the random effects regression, as the Hausmann Test is significant.
21
term and thus confirm the rejection of H1. Nevertheless, importantly, as opposed to the fixed effects regression, both R&D and multinationality are significant and positive. This is in line with existing theoretical and empirical evidence and suggests that cross-sectional effects might be more relevant than time-series effects within a firm when investigating the impact of these variables on cash holdings. The finding also helps explain the stagnating cash holdings in the UK, as the firms’ R&D expenditures decline and prevent an increase in line with the trend of other countries.
Second, it has been questioned whether the assumption of linearity holds for the relation between multinationality and the firms’ financial decisions (Mansi and Reeb, 2002). To test for possible non-linearity, the square of foreign sales to total sales is used as a robustness control. Indeed, the coefficient for this proxy is positive and significant at the 1% level, indicating that, given non-linearity, as firms reach high levels of internationalization they increase their cash holdings (Ramirez and Tadesse, 2009). The linear multinationality term is also significant in this regression, but with a negative coefficient. This means that firms increasing multinationality, but on a lower total level actually decrease their cash holdings. That being said, since any inclusion of quadratic terms into the interaction term does not yield different results from the linear terms regarding significance or direction, the non-rejection of the null-hypothesis is robust to these effects. The results are reported in Appendix C.
Further investigating this impact pattern, it is tested if either it might be more relevant whether a firm is a multinational or a domestic firm than to how large of a degree it has internationalized its sales or if it might be more relevant whether a firm does enact in R&D activities at all than to what degree it does so. Thus, dummy variables are added, which, regarding multinationality, equal 1 if a firm has more than 20% of total sales as foreign sales and 0 if it has 20% or less (Michel and Shaked, 1986) and, regarding R&D, equal 1 if a firm has R&D expenditures greater than zero and 0 if it does not. These dummy variables are substituting the main independent variables in the main fixed effects regression and very comparable, non-significant results are found and displayed in Appendix D.
22
shown in Appendix E. Multinationality is here positive and significant, indicating firms increasing their cash holdings in general before becoming multinationals. However, again the interaction terms are not significant, rejecting a stronger relationship between R&D and multinationality for firms in this state compared to other firms.
Additionally, four fixed effects quantile regressions are performed, each for domestic firms and for firms in different multinationality-tertiles to trace the impact of R&D upon cash holdings at various levels of multinationality. The variables are the same, except that the interaction term and the multinationality variable have been dropped, since the impact of multinationality is captured through the quantiles. As displayed in Appendix F, R&D is found to be insignificant for all four regressions, providing further robustness to the main regression.
Finally, a Variance Inflation Factor (VIF) test, the results of which are not reported here, is performed to test for possible multicollinearity among the variables. All values for VIF are smaller than 10, suggesting that none of the variables should be dropped in the regressions.
5. Conclusion
While Pinkowitz, Stulz and Williamson (2012) find a stronger relationship between cash holdings and R&D for multinationals compared to domestic firms in the US, my tests fail to replicate these findings for the theoretically similar UK. Neither the interaction term, nor the individual variables R&D and multinationality are significantly influencing cash holdings in the fixed effects regression. The interaction term and multinationality are also insignificant in the random effects regression. The results regarding the interaction term are robust to testing for non-linearity, the behavior as multinational compared to domestic firms and high-tech compared to non-high-tech firms as opposed to the degree of multinationality and R&D intensity using dummies, the impact of R&D on cash holdings of firms prior to becoming multinationals, multinationality-quantiles and multicollinearity.
23
and positive multinationality term in the robustness test looking at the three years prior to firms transitioning from a domestic to a multinational state. This regression is taking into account the findings of Pinkowitz, Stulz and Williamson (2012) that firms do not increase cash holdings at the point or after becoming multinationals, but instead before doing so. The findings are supported by the robustness test for non-linearity, which indicates that firms with an already high level of international sales do increase their cash holdings more with further internationalization of sales and, given a lower initial level, less so.
This has important implications as it highlights the non-generalizability of the results found in the US, which is critical both for academics researching cash holdings and for managers deciding on the optimal level of cash holdings in their firm when anticipating an increase of relative foreign sales. Possible reasons for the non-existence of a stronger relationship between R&D and cash holdings for multinationals, include abnormal country-specific trends in the UK and the relationship being a US-country-specific phenomenon.
The paper has two main limitations. First, regarding the research design, I use cash holdings instead of abnormal cash holdings. While this does affect the comparability to some extent, most of what is predicted in the abnormal cash holdings model should be captured by the control variables employed in this model. Second, cash holdings also depend on the intensity of competition and industry structure (Lyandres and Palazzo, 2012), as well as technology spillovers and product market rivalry (Qiu and Wan, 2012). Also, the findings of Zhou (2009), that cash holdings developed historically very differently in high-tech compared to non-high-tech sectors, suggest the importance of industry differences. Incorporating an industry dummy variable might therefore add to the explanatory power of the model.
24
the exact reasons why firms’ decisions on the amount of cash holdings in the 2000s in the UK differ from those in the US.
References
Almeida, H., Campello, M., & Weisbach, M. S. (2004). The cash flow sensitivity of cash. The Journal of Finance, 59(4), 1777-1804.
Almeida, H., Hsu, P. H., & Li, D. (2013). Less is More: Financial Constraints and Innovative Efficiency. Available at SSRN 1831786.
Arrow, K. (1962). Economic welfare and the allocation of resources for invention. In The rate and direction of inventive activity: Economic and social factors (pp. 609-626). Nber.
Bates, T. W., Kahle, K. M., & Stulz, R. M. (2009). Why do US firms hold so much more cash than they used to?. The Journal of Finance, 64(5), 1985-2021.
Baum, C. F., Caglayan, M., & Talavera, O. (2012). The effects of future capital investment and R&D expenditures on firms’ liquidity. Boston College Department of Economics. (No. 712).
Baysinger, B., & Hoskisson, R. E. (1989). Diversification strategy and R&D intensity in multiproduct firms. Academy of Management Journal, 32(2), 310-332.
Bell, A., & Jones, K. (2012). Explaining fixed effects: Random effects modelling of time-series cross-sectional and panel data. Centre for Multilevel Modeling, University of Bristol.
Black, F., & Scholes, M. (1973). The pricing of options and corporate liabilities. The Journal of Political Economy, 637-654.
25
Brown, J. R., & Petersen, B. C. (2011). Cash holdings and R&D smoothing. Journal of Corporate Finance, 17(3), 694-709.
De Jong, A., Huijgen, C. A., Marra, T. A., & Roosenboom, P. (2012). Why Do Firms Go Public? The Role of the Product Market. Journal of Business Finance & Accounting, 39(1‐2),
165-192.
Denis, D. J., & Sibilkov, V. (2010). Financial constraints, investment, and the value of cash holdings. Review of Financial Studies, 23(1), 247-269.
Dittmar, A., & Duchin, R. (2010). The dynamics of cash. Ross School of Business Paper, (1138).
Dittmar, A., & Mahrt-Smith, J. (2007). Corporate governance and the value of cash holdings. Journal of Financial Economics, 83(3), 599-634.
Dittmar, A., Mahrt-Smith, J., & Servaes, H. (2003). International corporate governance and corporate cash holdings. Journal of Financial and Quantitative analysis, 38(1), 111-134.
Duchin, R. (2010). Cash holdings and corporate diversification. The Journal of Finance, 65(3), 955-992.
Fama, E. F., & French, K. R. (2001). Disappearing dividends: changing firm characteristics or lower propensity to pay?. Journal of Financial economics, 60(1), 3-43.
Fama, E. F., & MacBeth, J. D. (1973). Risk, return, and equilibrium: Empirical tests. The Journal of Political Economy, 607-636.
Faulkender, M., & Wang, R. (2006). Corporate financial policy and the value of cash. The Journal of Finance, 61(4), 1957-1990.
26
Filatotchev, I., & Piesse, J. (2009). R&D, internationalization and growth of newly listed firms: European evidence. Journal of International Business Studies, 40(8), 1260-1276.
Foley, C. Fritz, Hartzell, J. C., Titman, S., & Twite, G. (2007). Why do firms hold so much cash? A tax-based explanation. Journal of Financial Economics, 86(3), 579-607.
Hall, B. H. (2002). The financing of research and development. Oxford review of economic policy, 18(1), 35-51.
Hall, B. H., & Lerner, J. (2009). The financing of R&D and innovation. National Bureau of Economic Research (No. w15325).
Himmelberg, C. P., & Petersen, B. C. (1994). R & D and internal finance: A panel study of small firms in high-tech industries. The Review of Economics and Statistics, 38-51.
Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. The American Economic Review, 76(2), 323-329.
Kalcheva, I., & Lins, K. V. (2007). International evidence on cash holdings and expected managerial agency problems. Review of Financial Studies, 20(4), 1087-1112.
Kamien, M. I., & Schwartz, N. L. (1975). Market structure and innovation: a survey. Journal of economic literature, 13(1), 1-37.
Keefe, M., & Kieschnick, R. (2012). Time Variation in the Marginal Value of Firms’ Cash Holdings. Available at SSRN 2120902.
Keynes, J. M. (1937). The general theory of employment. The Quarterly Journal of Economics, 51(2), 209-223.
27
Lyandres, E., & Palazzo, B. (2012). Strategic cash holdings and R&D competition: Theory and evidence. Available at SSRN 2017222.
Mansi, S. A., & Reeb, D. M. (2002). Corporate diversification: what gets discounted?. The Journal of Finance, 57(5), 2167-2183.
McLean, R. D. (2011). Share issuance and cash savings. Journal of Financial Economics, 99(3), 693-715.
Merton, R. C. (1973). Theory of rational option pricing. The Bell Journal of Economics and Management Science, 141-183.
Michel, A., and Shaked, I. (1986). Multinational corporations vs. domestic corporations: Financial performance and characteristics. Journal of International Business Studies, 89-100.
Miller, M. H., & Modigliani, F. (1961). Dividend policy, growth, and the valuation of shares. The Journal of Business, 34(4), 411-433.
Modigliani, F., & Miller, M. H. (1958). The cost of capital, corporation finance and the theory of investment. The American economic review, 48(3), 261-297.
Morck, R., & Yeung, B. (1991). Why investors value multinationality. Journal of Business, 165-187.
Mulligan, C. B. (1997). Scale economies, the value of time, and the demand for money: Longitudinal evidence from firms. Journal of Political Economy, 105(5), 1061-1079.
Myers, S. C. (1977). Determinants of corporate borrowing. Journal of financial economics, 5(2), 147-175.
28
Pinkowitz, L., Stulz, R., & Williamson, R. (2006). Does the contribution of corporate cash holdings and dividends to firm value depend on governance? A cross‐country analysis. The
Journal of Finance, 61(6), 2725-2751.
Pinkowitz, L., Stulz, R., & Williamson, R. (2013). Is There a US High Cash Holdings Puzzle after the Financial Crisis?. Fisher College of Business Working Paper, (2013-03), 07.
Pinkowitz, L., Stulz, R. M., & Williamson, R. (2012). Multinationals and the high cash holdings puzzle. National Bureau of Economic Research (No. w18120).
Qiu, J., & Wan, C. (2012). Technology Spillovers, Product Market Rivalry and Financial Flexibility. Product Market Rivalry and Financial Flexibility (November 1, 2012).
Ramirez, A., & Tadesse, S. (2009). Corporate cash holdings, uncertainty avoidance, and the multinationality of firms. International Business Review, 18(4), 387-403.
Ross, S. A. (1989). Institutional markets, financial marketing, and financial innovation. The Journal of Finance, 44(3), 541-556.
Schroth, E., & Szalay, D. (2010). Cash breeds success: The role of financing constraints in patent races. Review of Finance, 14(1), 73-118.
Shin, M. S., & Kim, S. E. (2011a). The effects of cash holdings on R&D smoothing: evidence from Korea. Journal of Finance & Accountancy, 6, 1-19.
Shin, M. S., & Kim, S. E. (2011b). The effects of cash holdings on R&D smoothing of innovative small and medium sized enterprises. Asian Journal of Technology Innovation, 19(2), 169-183.
Stiglitz, J. E., & Weiss, A. (1981). Credit rationing in markets with imperfect information. The American economic review, 71(3), 393-410.
29
Von Eije, H. (2012). What Causes Differences in Cash Holdings between Listed and Unlisted Firms around the World?. Available at SSRN 2010053.
Westerman, W. & Von Eije, H. (2005). Multinational Cash Management in Europe Towards Centralisation and Disintermediation: The Philips Case. Managerial Finance, 31(10), 65-74.
30 Appendix A Explanation of the Variables
Name indicates the name of the variable. Variable is the items taken from datastream to calculate it. Use is its function within the regressions. Exp. Sign is the expected direction of impact, i.e. the sign of the coefficient. TA is total assets. OI is operating income. MVE is market value of equity. BVE is book value of equity. CAPEX is capital expenditures. TD is total debt. LN is the natural logarithm.
Name Variable Use Exp. Sign
CH Cash (+equivalents) / TA Dependent variable RD*
MN
(R&D expenses / TA) * (Foreign sales / total sales)
Interaction effect – main independent +
MN Foreign sales / total sales Multinationality – main independent + RD R&D expenses / TA R&D - main independent +
RI SD (OI
t / TAt; OIt-1 / TAt-1; OIt-2 /
TAt-2)
Risk of cash flows +
GO (TA + MVE – BVE / TA) Growth opportunities +
CF OI / TA Cash flows +
CE CAPEX / TA Capital expenditure -
LE TD / TA Leverage -
LQ Receivables + Inventory / TA Liquidity -
SL Short-term debt / TA Short-term access to liquidity -
FS LN(TA) Firm size -
FA Date of incorporation Firm age -
31 Appendix B Fama-MacBeth Regressions
Results for the Fama-MacBeth regressions for the UK. The dependent variable is cash holdings. T indicates the t-value and P the level of significance. * indicates a significance level of 10%, ** of 5% and *** of 1%. Year indicates the year and Observ. the corresponding observations. RD*MN is the interaction variable of research and development and multinationality. MN is multinationality. RD is research and development. RI is risk. GO is growth opportunities. CF is cash flow. CE is capital expenditure. LE is leverage. LQ is liquidity. SL is short-term access to liquidity. FS is firm size. FA is firm age. DI is dividends. C is the coefficient. The number of yearly observations is between 268 in year 2000 and 356 in year 2007. All financial data are calculated as ratios to the firm’s total assets. All variables except for firm size and firm age are winsorized at the 1st and 99th percentiles. All variables except for the dependent variable cash holdings and firm age are lagged by one year.
32 Appendix C Non-linearity
Results for the panel regression with fixed effects for the full sample of firms from the UK for 2000-2012 . The dependent variable is cash holdings. C indicates the coefficient of the variable and P the level of significance. *
indicates a significance level of 10%, ** of 5% and *** of 1%. RD*MN² is the squared interaction variable of
research and development and multinationality. RD*MN is the interaction variable of research and development
and multinationality. MN² is squared multinationality. MN is multinationality. RD is research and development.
RI is risk. GO is growth opportunities. CF is cash flow. CE is capital expenditure. LE is leverage. LQ is liquidity. SL is short-term access to liquidity. FS is firm size. FA is firm age. DI is dividends. C is the coefficient. Observ. is the total amount of observations for which all information was available. Adj. R² is the adjusted R-squared. Prob. indicates the significance of the equation. All financial data are calculated as ratios to the firm’s total assets. All variables except for firm size and firm age are winsorized at the 1st and 99th percentiles. All variables except for the dependent variable cash holdings and firm age are lagged by one year. Time dummies for the years 2002-2012 have been added, but are not shown.
33
Appendix D Dummy variables for main independent variables
Results for the panel regressions with fixed effects for the full sample of firms from the UK for 2000-2012. The dependent variable is cash holdings. C indicates the coefficient of the variable and P the level of significance. * indicates a significance level of 10%, ** of 5% and *** of 1%. DRD*DMN is the interaction variable of the dummies of research and development (1 if > 0, 0 otherwise) and multinationality (1 if > 0.2, 0 otherwise). DMN is the dummy of multinationality. DRD is the dummy of research and development. RI is risk. GO is growth opportunities. CF is cash flow. CE is capital expenditure. LE is leverage. LQ is liquidity. SL is short-term access to liquidity. FS is firm size. FA is firm age. DI is dividends. C is the coefficient. Observ. is the total amount of observations for which all information was available. Adj. R² is the adjusted R-squared. Prob. indicates the significance of the equation. All financial data are calculated as ratios to the firm’s total assets. All variables except for firm size and firm age are winsorized at the 1st and 99th percentiles. All variables except for the dependent variable cash holdings and firm age are lagged by one year.
34 Appendix E Pre-multinational dummy
Results for the panel regressions with fixed effects for the full sample of firms from the UK for 2003-2012 . The dependent variable is cash holdings. C indicates the coefficient of the variable and P the level of significance. * indicates a significance level of 10%, ** of 5% and *** of 1%. RD*MN is the interaction variable of research and development and multinationality. MN is multinationality. RD*DPMN is the interaction variable of research and development and a dummy for multinationality, that takes 1 for t-3, t-2 and t-1 when a firm reaches a higher level than 20% of foreign sales to total sales, and 0 if it does not. DPMN is the dummy for multinationality. RD is research and development. RI is risk. GO is growth opportunities. CF is cash flow. CE is capital expenditure. LE is leverage. LQ is liquidity. SL is short-term access to liquidity. FS is firm size. FA is firm age. DI is dividends. C is the coefficient. Observ. is the total amount of observations for which all information was available. Adj. R² is the adjusted R-squared. Prob. indicates the significance of the equation. All financial data are calculated as ratios to the firm’s total assets. All variables except for firm size and firm age are winsorized at the 1st and 99th percentiles. All variables except for the dependent variable cash holdings and firm age are lagged by one year.
35 Appendix F Quantile Regressions
Results for the quantile panel regressions with fixed effects for domestic firms (foreign sales / total sales = 0) and tertiles of multinationality from the UK for 2000-2012 . MN-Tertile 1 is the smallest multinationality tertile of firms with foreign sales, MN-Tertile 2 is the mid multinationality tertile and MN-Tertile 3 is the biggest multinationality tertile. The dependent variable is cash holdings. C indicates the coefficient of the variable and P the level of significance. * indicates a significance level of 10%, ** of 5% and *** of 1%. RD is research and development. RI is risk. GO is growth opportunities. CF is cash flow. CE is capital expenditure. LE is leverage. LQ is liquidity. SL is short-term access to liquidity. FS is firm size. FA is firm age. DI is dividends. C is the coefficient. Observ. is the total amount of observations for which all information was available. Adj. R² is the adjusted R-squared. Prob. indicates the significance of the equation. All financial data are calculated as ratios to the firm’s total assets. All variables except for firm size and firm age are winsorized at the 1st and 99th percentiles. All variables except for the dependent variable cash holdings and firm age are lagged by one year.
Domestic Firms MN-Tertile 1 MN-Tertile 2 MN-Tertile 3
