The Cash Paradox: The effect of Language on
Corporate Cash Holdings
by
TOM DOKTER
University Of Groningen
Faculty of Economics and Business
Msc International Financial Management
June 2015
Field Key Words: Cash Holdings, Research and Development, Cash Paradox,
Culture, Agency Theory, Determinants, Language
Supervisor:
Lamers, M.A.
Van Disselstraat 11
8044 PL Zwolle
(06) 55154911
tom_dokter@hotmail.com
student number: s2033224
Abstract: This study examines the relationship between the native language of executives and corporate cash holdings. Thereby, this paper aims to explain the variation in corporate cash levels. We find that firms whose executives natively speak languages that do not grammatically differentiate between the future and present have statistically significant lower levels of corporate cash holdings compared to executives that do grammatically differentiate future from present. The amount of differentiation is called future time reference (FTR). Hence, there is a negative relationship between FTR and the level of corporate cash holdings. Additionally, this study finds that investment size is positively related to the level of corporate cash holdings. As industries are becoming increasingly high-tech, this study provides statistical evidence for the rise in cash holdings over the past decades.
Content
1. Introduction ... 3
2. Literature Review ... 5
3. Data & Methodology ... 8
1. Introduction
Corporate cash holdings have grown significantly over the past decades (Amess, Banerji and Lampousis, 2015). More specifically, the 2014’s report by Deloitte illustrates the top 1000 non-financial organizations in the world hold 2.8 trillion US dollars in cash. More interestingly, Deloitte reports that only 25% of the non-financial FTSE 100 firms account for 80% of these global cash holdings (Deloitte, 2014). To illustrate this phenomenon occurs globally consider the case for Australia where only 20% of the ASX 200 companies account for 82% of the country’s total cash reserves (Deloitte, 2015).However, these companies have all been underperforming in quarterly revenue growth or share price performance vis-à-vis their competitors with relatively small cash holdings (Deloitte, 2014). This implies there is a cash paradox, which increasingly interests academic scholars (Amess et al, 2015).
Hitherto, literature illustrates a broad range of determinants for cash holdings on the firm level. These determinants can be divided amongst 4 different motives for holding cash (Bates, Kahle, and Stulz, 2009). First of all, is the transaction motive, where the optimal level of cash is determined by the transaction costs incurred to convert non-cash assets into cash (Miller and Orr, 1966). Moreover, due to economies of scale exist larger firms hold less cash according to this motive (Bates et al, 2009). Secondly, is the precautionary motive, which states that firms hold additional cash in order to safeguard against future economic shocks (Bates et al, 2009). This motive also states that firms with more volatile cash flows, better investment opportunities, and poor access to external capital hold more cash (Almeide, Campello, and Weisbach, 2004; Han and Qiu, 2007). Thirdy, there is a tax motive. According to Foley, Hartzell, Titman, and Twite (2007) MNEs hold higher levels of cash as these firms incur tax high tax costs when the income of foreign affiliates is returned to the corporate headquarters. Finally, the agency motive predicts that managers have an incentive to hold higher levels of cash as they have the ability to use cash at their own discretion (Chang and Noorbakhsh, 2009). Dittmar, Marth-Smith, Servaes (2003) used international firm data to test the correlation between cash holdings and shareholder protection, and found that insufficient shareholder protection increases managers’ preferences to hold additional cash. Hence, their result support the agency theory and illustrate that a high level of cash holdings can be seen as evidence for an agency problem within that organization.
However, prior studies do not provide a significant explanation for the rise in cash holdings over the past decades. Additionally, Amess, Banerji, and Lampousis (2015) state that the global improvements of technologies permit advanced inventory management and improved access to capital, which would reduce the need for cash Hence, it is unclear why organizations still hold increasing amounts of cash. To address this issue most of the current literature emphasises on precautionary and agency motives for an organization’s cash holding. Whilst the results do partly explain the variation amongst organizations, they
do not adequately explain the before mentioned increase in cash holdings. Therefore, to better understand the increase in organizations’ cash holdings Amess et al. (2015) suggest several headings for future research through the review of prior studies on cash holdings. First of all, they illustrate the lack of emphasis on cultural factors in prior studies as only recently financial studies have taken an interest in the role of culture on cash holdings. The importance of cultural factors is by Chang and Noorbakhsh (2009) who conclude that corporate cash holdings are influenced by the national cultures in which these firms operate. Additionally, Chen, Dou, Rhee, Truong, and Veeraragavan (2015) find similar results and conclude that culture can be used to explain the variation in cash holdings around the world. Thus, studies that do not include cultural variables might be biased as coefficients of other variables are correlated with the omitted cultural variables. Secondly, Amess et al. (2015) suggest the rocketing increase in investment costs in some industries might have increased precautionary cash holdings. For example, high-tech industries find it difficult to obtain external funding for their investments and therefore primarily rely on cash to fund investments (Himmelberg and Peterson, 1994). Additionally, it is commonly accepted that the global economy is becoming increasingly high-tech. Hence, we might expect MNEs to hold higher levels of cash as a precaution. Moreover, by closely examining the trend in R&D expenditures of the global top 1000 public companies, PWC finds that R&D spending has increased by 5.1% in 2015 (Jaruzelski, Schwartz, Staack, 2015). Additionally, they conclude that companies continuously increase their R&D expenditures even though revenues tend be more volatile over the past decade.
By employing the headings offered by Amess et al. (2004) aims to provide a contribution to the existing literature. In order to achieve this, our study uses the paper of Chen et al. (2015) as a background paper. However, unlike their paper and in line with Guin (2015) we argue that speaking a similar language can be seen as a cultural dimension and is a prerequisite for social interaction. More specifically, speaking a similar language enables the transmission of beliefs and preferences amongst its members. Besides, some languages grammatically connect present and future, whilst other languages grammatically separate present from future, which influences the economic behaviour of individuals. (Chen and Keith, 2013). Hence, we use language as a proxy to measure culture instead of the cultural factors used by Chen et al, (2015). Moreover, this study uses a sample of 731 listed firms within the Euro-zone between 2010-2014 to examine the relationship between corporate cash holdings and the native language of both CEO and CFO.
Therefore, this study will contribute to the existing literature in the following ways. First of all, this study provides additional evidence for the cultural motive as the results show that executives’ national culture is an important determinant of corporate cash holdings. Hence, it is shown that corporate decision making is also dependent on executives’ subjective beliefs. Secondly, we demonstrate that there is an relationship
between speaking certain languages and the level of cash holdings. By examining this effect on a corporate level, we provide new insight in the role of languages.
The remainder of this paper begins with a review of the existing literature in section 2. The next section describes the data, variables and control measures used, and provides descriptive statistics of our sample. Section 4 illustrates the results of our statistical test. And the next section discusses the implications of this study and provides directions for future research. Finally, in section 6 summarizes the research and present the conclusion.
2. Literature Review
One of the earlier studies on cash holdings by Myers and Majluf (1984) concludes that companies are more likely to hold extra cash in order to avoid the need of external funds in case of economic downfall. Kim, Mauer, and Sherman (1998) underline this by illustrating there is a trade-off in a company’s optimal cash holdings between the low returns generated by holding cash and ability to fund investments opportunities in the future. These opportunities would otherwise be funded by generally more costly external financing. Moreover, Opler, Pinkowitz, Stulz, and Williamson (1999) conclude that the level of cash holding have a positive relationship with both growth opportunities and the riskiness of the investments, and a negative relationship with access to capital. Besides that, size should have a negative effect on cash holdings. (Oppler et al, 1999; Ozkan and Ozkan, 2004) Moreover, younger and smaller companies hold cash as a precaution for volatile R&D expenditures. (Brown and Peterson, 2011)
One of the main take-aways from previous research is that ‘cash provides unconditional liquidity at all times
and states of the world’ whilst (Amess et al, 2015, p425) Therefore, it might be preferred over bank
financing. However, Pinkowitz and Williamson (2001) illustrate by examining German, US, and Japanese companies that cash holdings are significantly influenced by the monopoly power of banks. Hence, a company operating in a country where banks are relatively powerful might hold more cash than its competitors in other parts of the world. The reason for this is that powerful monopolistic banks can request firms to hold higher levels of cash as in order to receive financing (Pinkowitz and Williamson, 2001). Additionally, one might argue that these banks are able to demand higher interest from firms due to their strength, which increases a firms need for cash. This finding is partially confirmed by the results of Ozkan and Ozkan (2004) who find that lower levels of both bank debt and leverage result in higher corporate cash holdings in the UK. Hence, in the absence of powerful monopolistic banks lower bank debt and leverage results in higher corporate cash holdings.
(2010) find that excess cash has a positive relationship with to a company’s capital investments. Additionally, the study illustrates the more financially constrained a company is, the more it reduces its cash holdings. A more recent study by Bliss, Cheng, and Denis (2015) that examines the relationship between corporate payout and cash retention during the 2008 financial crisis, finds that payout reductions are larger during the crisis period. More interestingly, they find that the cash saved through these reductions have a positive relationship with both the level of investments and the cash balances during the crisis. These findings underline the notion that companies might increase their cash holdings to preserve investments. These results are underlined by findings during other financial crises around the globe. For example, Song and Lee (2012) find there are long term effects of a financial crisis on a company’s cash holding. More specifically, they find that public companies in countries affected by the Asian financial crisis in 1997-1998 lowered their investments and increased their level of cash in the long-term even after economic recovery. As information asymmetries are deemed rather important for R&D expenditures, companies with higher R&D expenses should therefore hold more liquid assets. (Oppler et al, 1999)
Hence, the first hypothesis can be stated as follows:
Hypothesis 1. Ceteris-paribus, the level of corporate cash holdings is positively related to an MNE’s investment size.
CULTURE
National culture can affect corporate decision making through two separate channels (Chen et al, 2015). First of all, national culture affects a manager’s view and preferences, which might affect corporate decision-making. Secondly, national culture affects investors’ views and preferences, which in turn affects corporate decision-making as firms try to align their policies with the views and preferences of potential investors (Chen et al, 2015). For the purpose of the paper we will emphasize on the prior channel, as our main interest is the effect of national culture on corporate policies, in our case cash holdings, on the board-level.
Throughout the existing literature, Hofstede’s framework, (1980, 1991, 2001) is used to analyze and measure the effect of culture on organizations and corporate policies. Hofstede defines culture as “the
collective programming of the mind which distinguishes the members of one human group from another”.
The initial framework consisted of four dimensions; the tolerance of power distance, individualism, masculinity versus femininity, and uncertainty avoidance. In later studies Hofstede (1991, 2001) added two dimensions; long term orientation versus short term orientation, and indulgence versus restraint.
countries where the national culture scores high on uncertainty avoidance, is more masculine, and is characterized by long term orientation. This is underlined by Chen et al. (2015) who measure culture with two factors from Hofstede (1980), which are uncertainty avoidance and individualism. Moreover, their study illustrates a negative relationship between cash holdings and individualism. Therefore, companies operating in more individualistic countries hold less cash compared to firms operating collectivistic countries. Additionally, they find a positive relationship for uncertainty avoidance. Hence, countries that score higher on the uncertainty avoidance dimension have higher levels of cash. This can be explained by the precautionary argument first mentioned by Myers and Majluf (1984), as companies from countries with higher uncertainty avoidance hold higher levels of cash in order to avoid the need of external funds in case of economic downfall.
However, Hofstede’s framework has been subject to criticism due to its simplistic conceptualisation of culture. (Kirkman et al. 2006) Moreover, doubts have been raised about the framework’s appropriateness to measure culture in the 21st century as the sample is limited to a single company and the framework is based
on data from the 1960s and the 1970s. Hence, another definition of culture is provided by North (1990) who defines culture as ‘transmission from one generation to the next, via teaching and imitation, of knowledge,
values, and other factors that influence behaviour.’ For the purpose of this paper we use North’s (1990)
definition of culture.
based on two different mechanisms. First of all, language can change how distant future event feel, hence weak-FTR speakers would perceive future events as less distant. Mezhevich (2008) underlines this by stating that speakers of a language that uses a future or past tense when speaking of current event, view future events as more distant. How this influences saving behaviour of individuals can be easily explained through the use of discount rates, or beliefs about the future. As noted prior speakers of weak-FTR languages perceive the future as less distant, hence, their perceived discount rate is lower compared to speakers of strong-FTR languages. More specifically, speakers of strong-FTR languages believe future rewards to be more distant, which creates an incentive to accept immediate rewards. On the contrary, speakers of weak-FTR languages perceive future rewards as less distant and are therefore more willing to save. Secondly, Chen (2013) argues that speakers of strong-FTR languages hold more precise beliefs about the timing of future events and thus future rewards. Moreover, ‘uncertainty about the timing of future payoffs makes saving more attractive’. (Chen, 2013, pp 698) This is underlined by the findings of Redelmeier and Heller (1993). Finally, the findings of Chen (2013) support this school of thought. Hence our two-part, hypothesis is;
Hyothesis 2(a). There is a relationship between the level of cash holdings at the corporate level and the native language of executives.
Hypothesis 2(b). Ceterus-paribus, the level of corporate cash holdings is higher for MNEs with executives from countries with weak FTR languages.
3. Data & Methodology
3.1 Data
are identified by their NACE code. It is important to note that the financial sector and other heavily state regulated industries e.g. healthcare were excluded from this research. As these firms are expected to be outliers in any study across industries (Cooper, Jackson, and Patterson, 2003) Moreover, the cash holdings of these firms is expected to be primarily influenced by capital requirements rather than economic reasons (Bates et al., 2009). Similarly, organizations that did not have information available on the nationality of either CEO or CFO were discarded. After applying the filters our final sample consists of a total of 3186 observations across 637 listed companies over a period of 5 years. The financial data is gathered from balance sheets and income statements available through the Orbis database. Finally, to account for the effect of outliers we winsorize the observations at the 1st and 99th percentile. The overall descriptive statistics of
our final sample are presented in Table 1, which depicts the amount of observations, mean values, median values, and both minimal and maximum values. The average and the median cash holdings of firms in our final sample, reported in the 2nd and 5th column of table 1, are 9.35% and 7.37% respectively. Additionally,
the standard deviation, reported in column 3, is around 7.6%, which implies that there is quite some variation in cash holdings within our sample. This is underlined by the information in the last two columns of table 1, where the minimum and maximum values are depicted, as the level of cash holdings ranges from 0.2% to 37.2%.
Table 1
Summary statistics.
This table reports the number of observations, means, standard deviations, medians, minimum, and maximum values of our sample. The sample consists of listed firms from the EURO-zone during the period 2010 to 2014. Cash holdings are measured as the ratio of cash and equivalents to total assets. Cash flow is the ratio of operating cash flow to total assets, i.e., ocf/ta. Size is the log of total assets, i.e., log(ta). Liquidity, is the net working capital minus cash to total assets, i.e., (nwc-cash)/ta. Investment size, is the log of R&D expenditures, i.e., log(rd). Strong FTR, is based on the nationality of CFO and CEO and a binary variable that indicates whether their native language grammatically separates future from present, where a value of 1 indicates that a language does and a value of 0 indicates the opposite.
Variable # Obs Mean St. dev Min Median Max
Cash holdings 3204 0.094 0.076 0.002 0.074 0.372 Cash flow 3204 0.060 0.067 -0.109 0.057 0.291 Size 3204 13.52 1.937 10.12 13.28 18.39 Liquidity 3204 0.059 0.160 -0.435 0.062 0.451 Leverage 3197 0.656 0.575 0.168 0.579 5.258 Investment size 1693 4.303 0.949 2.045 4.320 6.621 Strong FTR 3201 0.592 0.492 0 1 1
Moreover, the number of observations on investment size is relatively small compared to the rest of variables, which could significantly decrease our sample size when included in the analysis. However, we
still aim to use investment size as a variable in the analysis as prior studies illustrate it is a significant determinant of corporate cash holdings. Besides, Amess et al. (2015) indicated that the role of investment size may have become more important over the past decades due to the fact that the global economy is becoming increasingly high-tech. Hence, this study will use different models both including and excluding investment size in order to preserve the sample size and to ensure we include the effect of investment size in our analysis. Additionally, the summary statistics of the variables across the several countries are illustrated in Table 2. Similar to Chen et al. (2015) all numbers for the respective variables are median values except for the number of observations and firms per country. Furthermore, France and Germany have the largest representation in the sample with 710 and 656 observations respectively. However, this is to be expected as both countries are commonly considered to be the largest economies in the EURO-zone. Moreover, to ensure the statistical results are not influenced by these two countries, we test for robustness by excluding these markets in a later section.
Table 2
Summary statistics (per country)
This table illustrates the median values for all countries present in the sample. All entries are country median values with the exception of the number of observations. The sample consist of listed firms within the EURO-zone during the period of 2010 to 2014. Cash holdings are measured as the ratio of cash and equivalents to total assets. Cash flow is the ratio of operating cash flow to total assets, i.e., ocf/ta. Size is the log of total assets, i.e., log(ta). Liquidity, is the net working capital minus cash to total assets, i.e., (nwc-cash)/ta. Investment size, is the log of R&D expenditures, i.e., log(rd). Strong FTR, is based on the nationality of CFO and CEO and a binary variable that indicates whether their native language grammatically separates future from present, where a value of 1 indicates that a language does and a value of 0 indicates the opposite. n.a. indicates the median value for the respective variable is not available.
Country # Obs. # Firms Cash holdings Cash flow Size Liquidity Leverage Investment size
Additionally, the median values indicate there are 6 countries with weak FTR languages and 12 countries with strong FTR languages in our sample. However, it must be stated that variation exists within countries as, some executives are expats and therefore considered to be speakers of a foreign language. The overall mean values of cash holdings in strong-FTR countries and weak-FTR countries are 5.32% and 7.20% respectively. This may already indicate that speakers of weak-FTR languages hold higher levels of cash. However, to statistically infer that the median cash holdings are higher for speakers of weak-FTR languages than for strong-FTR languages a one tailed t-test can be used to compare the two separate groups. (Newbold, Carlson, and Thorne, 2012) This test provides a t-statistic of 1.45 and provides significant univariate evidence at the 10% level that the median values for cash holdings are higher for strong-FTR languages. The distribution of the languages per company per country are illustrated in table 4 also underlines the notion that there may not be enough variation in FTR within the Euro-zone. Similar to the country statistics the French and German language have the largest representation in our sample with 143 and 154 executives respectively. Additionally, the variation within markets in rather small as most countries are dominated by native speakers of the domestic language. Some clear examples are Finland, France, Germany, Greece, and Italy. In these 5 countries combined only 9 executives out of 458 do not have the local nationality and thus do not have the FTR-score for the local language. On the other hand, our sample for the Euro-zone contains 20 different languages. However, most of the languages have relatively little observations, e.g. Chinese and Estonian with 2 observations each.
Table 4 illustrates the median values across the three sectors of interest; primary, manufacturing, and services. It is important to note that the manufacturing sector with 2260 observation is the largest in our sample due to the exclusion e.g. health care, the financial sector and other heavily regulated service industries. The highest median cash holdings are in the service sector with a value of 7.6% of total assets, which is relatively similar to the manufacturing sector with a median value of 7.3% of total assets. On the contrary, the primary sector illustrates there is a difference between the separate sectors as it has a median value of 6.1% of total assets. These median values correspond to our findings for the average values for strong-FTR in the three subsamples. Basically, the primary sector scores the highest for strong-FTR with 77.9% of the executives speaking strong-FTR languages, whilst in both the manufacturing and services have relatively lower and similar values with 58.5% and 57.8% respectively. It is important to remember we hypothesised companies with executives that speak weak-FTR languages hold higher levels of cash, which is underlined by these preliminary findings. More specifically, the lowest score on strong-FTR (57.8%) results in the highest score for cash holdings (7.6%).
Table 4
Language distribution (per country)
This table illustrates the language distribution all countries present in the sample. All entries are number of firms with executives who speak a specific language in a specific country with the exception of the total number of observations per language and FTR-scores per language. E.g. in the first row Chinese has an FTR score of 0 (weak-FTR) and there are 2 executives speaking Chinese. In this case both operate in Germany. The sample consist of listed firms within the EURO-zone during the period of 2010 to 2014. FTR, is based on the nationality of CFO and CEO and a binary variable that indicates whether their native language grammatically separates future from present, where a value of 1 indicates that a language does and a value of 0 indicates the opposite.
Language FTR #Obs Austria Belgium Cyprus Estonia Finland France Germany Greece Ireland Italy Lithuania Luxembourg Malta Netherlands Portugal Slovakia Slovenia Spain
Other notable differences are: the differences in cash flow, where the median cash flow for the primary sector is the highest with 6.8%, and the difference in leverage, where the services sector has the highest median leverage with 63.2%. On the latter variable, both primary and manufacturing sector have a median value of around 56%. Hence, we will control for industry fixed effects by adding industry dummies to our regressions. Furthermore, we will do a more elaborate sectoral analysis in section 7 to illustrate the impact of language per sector and to examine the variation of our hypothesized relationship between the three selected sectors.
The correlation coefficients between the variables used for our analysis are presented in Table 5. In line with our main hypothesis, these correlations illustrate the expected negative relationship between our language variable and cash holdings. However, we need more evidence to support our hypothesis as other controlled effects are absent in this representation. Additionally, there is no sign of multicollinearity between the variables.
3.2 Empirical Model
In order to analyse the cash holdings of firms within the Euro-zone our empirical model will combine the models of prior studies (e.g., Ozkan and Ozkan, 2004; Bates et al, 2009; Chang and Noorbakhsh, 2009; Chen and Keith, 2013; Chen et al, 2015). Therefore the framework includes the following variables: cash flows, liquid assets, leverage, size, culture, investment size. Through conducting a Hausman test we find that the our model cannot adequately estimated using random-effects. However, as our language variable remains constant during the period we are not able to use panel data with fixed effects. Remember that we assumed that the executives did not change during our period of interest. Besides that, Dahl (2000) illustrates that languages are relatively stable over time, e.g. Germanic languages do not grammatically separate future from present since approximately 2,000 years ago. Therefore, in line with previous cultural studies, we use pooled OLS to run the regressions, where we control for time specific effects (Chang and Noorbakhsh, 2009; Chen et al, 2015). Moreover, all firms within our sample hold cash and equivalents. Hence, we do not first test for the presence of cash holdings, but start to test whether the differing levels of cash can be explained by the language spoken.
The model is:
𝐶𝐶𝐶𝐶𝐶𝐶ℎ𝑖𝑖𝑖𝑖 =∝ + 𝛽𝛽1𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖+ 𝛽𝛽4𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖+ 𝛽𝛽5𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖𝑖𝑖+ 𝛽𝛽7𝑆𝑆𝐿𝐿𝑆𝑆𝐿𝐿𝑖𝑖𝑖𝑖+ 𝛽𝛽8𝐼𝐼𝐼𝐼𝐿𝐿𝑆𝑆𝐿𝐿𝑆𝑆𝐿𝐿𝑖𝑖𝑖𝑖+ 𝛽𝛽9𝐿𝐿𝐶𝐶𝐼𝐼𝐿𝐿 + 𝑌𝑌𝑖𝑖𝑖𝑖 + 𝜀𝜀𝑖𝑖𝑖𝑖
In this model i represent the company, and t represents the year. Cashit is an MNE’s cash holding and is the
dependent variable of this study. To measure this variable this research follows the method of Chang and Noorbakhsh (2009) and other existing literature.
Table 3
Summary statistics (per sector)
This table illustrates the median values for the three separate sectors present in the sample. The companies in the sample are categorized according to their NACE-industry codes. All entries are sector median values with the exception of the number of observations, the number of firms, and the mean values for Strong FTR. The sample consist of listed firms within the EURO-zone during the period of 2010 to 2014. Cash holdings are measured as the ratio of cash and equivalents to total assets. Cash flow is the ratio of operating cash flow to total assets, i.e., ocf/ta. Size is the log of total assets, i.e., log(ta). Liquidity, is the net working capital minus cash to total assets, i.e., (nwc-cash)/ta. Investment size, is the log of R&D expenditures, i.e., log(rd). Strong FTR, is based on the nationality of CFO and CEO and a binary variable that indicates whether their native language grammatically separates future from present, where a value of 1 indicates that a language does and a value of 0 indicates the opposite.
Sector # Obs. # Firms Cash holdings Cash flow Size Liquidity Leverage Investment size Strong FTR
Primary 145 29 0.061 0.068 14.60 0.114 0.564 4.185 0.779
Manufacturing 2260 452 0.073 0.060 13.27 0.074 0.567 4.356 0.585
Services 743 149 0.076 0.047 13.25 -0.006 0.632 3.917 0.578
Table 4
Correlation coefficients
The sample consist of listed firms within the EURO-zone during the period of 2010 to 2014. Cash holdings are measured as the ratio of cash and equivalents to total assets. Cash flow is the ratio of operating cash flow to total assets, i.e., ocf/ta. Size is the log of total assets, i.e., log(ta). Liquidity, is the net working capital minus cash to total assets, i.e., (nwc-cash)/ta. Investment size, is the log of R&D expenditures, i.e., log(rd). Strong FTR, is based on the nationality of CFO and CEO and a binary variable that indicates whether their native language grammatically separates future from present, where a value of 1 indicates that a language does and a value of 0 indicates the opposite.
Cash holdings Cash flow Size Liquidity Leverage Strong FTR
Hence, we use the ratio of cash and short term investments to net assets. In this case net assets is total assets minus cash and equivalents. Data on cash and short term investments is widely available through the Orbis database. Our main explanatory variable is Langit and is based on the framework by Chen (2013). Hence,
language is a dummy variable to separate speakers of strong FTR languages from speakers of weak FTR languages. As indicated earlier, this study emphasises on the effect of the language spoken by executives on corporate cash holdings. This implies that the CEO or CFO from a specific company is given a score of either 1 in case he or she speaks a strong-FTR language and a score of 0 if he or she speak a weak-FTR language. Furthermore, several control variables are included. First of all, CFlowit is the operating cash flow.
Similarly to prior studies this variable is measured as a ratio of operating cash flow to total assets. (Ozkan and Ozkan, 2004; Chen et al., 2015) Data for this variable is gathered from the Orbis database. Secondly,
Liqit stands for non-cash liquid assets. Like prior research, to measure this variable the ratio of net working
capital minus cash to total assets is used as a proxy. (Ozkan and Ozkan, 2004) Additionally, Levit is the
amount of financial leverage in the specific organization. Leverage is measured by the debt to assets ratio, which is commonly defined as the ratio of long term debt plus short term debt to the book value of assets. (Ross, Westerfield, Jaffe, and Bradford, 2011) We also control for size of the firm, sizeit , which is proxied
by the log of total assets (Opler et al, 1999) As illustrated earlier in this study, large companies tend to have less information asymmetries than small firms. Moreover, size should have a negative effect on cash holdings. InvSizeit represents the absolute size investments. To measure this I will use the log of R&D
expenditures as a proxy. As illustrated previously it is generally accepted that the global economy is increasingly moving towards high-tech industries. Therefore, investment size increases, which results in higher cash holdings. Finally, Yit is a dummy variable to control for time specific effects and 𝜀𝜀𝑖𝑖𝑖𝑖 is the
standard error term.
4. Results
In this section we present the results of our linear regressions. First of all, we conduct univariate regressions to illustrate whether language has an effect on corporate cash holdings. Afterwards, we include control variables. The regression output is shown in table 5. The first column of this table depicts the results for the full sample without control variables. Model 2 includes control variables and the third model also includes investment size. The output of model 1 illustrates a significant negative relationship between FTR and corporate cash holdings at the one percent level. More importantly, the negative relationship remains statistically significant at the one percent level after the inclusion of controls. As noted earlier, the sample size significantly decreases when investment size is included. However, it is important to state that the relationship stays statistically negative, which indicates that speakers of weak-FTR languages hold higher
levels of cash. These findings are consistent with hypothesis 2 that ceteris-paribus, the level of corporate cash holdings is higher in firms if executives natively speak weak-FTR languages.
Moreover, the magnitude of coefficients increase, from -.019 to -.015 to -.008, indicating the effect becomes smaller. The coefficient of -.015, in the second column, indicates that ceteris-paribus firms with executives from countries with strong FTR languages have a between 0.8% and 1.5% lower cash ratio than their weak FTR counterparts.
Investment size illustrates the hypothesized positive effect at the one-percent level. The coefficient is .021 in model 3, which implies that a 1% increase in the size of investments increases the cash holdings by 2.1 percentage points. Model 4 to 6 in table 5 excludes France and Germany from the sample due to their large number of observations compared to other countries. The output underlines that investment size is positively correlated to cash holdings with a statistically significant positive coefficient, .024, at the one-percent level in model 6. These findings are in line our first hypothesis that investment size has a positive relationship with cash holdings.
Furthermore, model 4 and 5 indicate that FTR becomes statistically insignificant after extracting France and Germany in. Additionally, the coefficients in these models imply a positive relationship between FTR and corporate cash holdings, whilst we hypothesized a negative effect. Even though model 4 and 5 are not statistically significant, model 6 illustrates this positive relationship is statistically significant at the five-percent level after the inclusion of all control variables. This indicates that firms with executives from countries with strong FTR languages hold higher levels of cash. Thus the results might be determined by France and Germany. However, it is important to note that the sample size is approximately halved compared to the first three models. Besides, the most of the control variables are insignificant in model 5 and 6. More specifically, only the cash flow illustrates the expected positive relationship and is statistically significant at the one-percent level in both models, whilst the full-sample models confirms the expected signs for all variables at the one-percent level. The coefficient of .201 in model 5 implies that a 1 percentage-point increase in cash flow to net assets leads to a 20.1% increase in cash holdings. Additionally, the adjusted R2 is higher for the full sample models, 1 to 3, implying that in these models the proportion of sample
variance explained by the variables is greater (Brooks, 2014). Model 7 to 9 focus only use the observations from France and Germany. The output shows that the expected negative coefficients between FTR and cash are relatively larger than for the full-sample. More specifically, the negative effect in model 7 to 9 ranges from -.036 to -.025, which indicates that executives who speak strong-FTR languages have a between 2.5% and 3.6% lower cash ratio in their firms than executives who speak weak-FTR languages. If we compare this to the results of the full-sample models, it becomes clear that the effect is more than twice as large. This underlines the notion that both France and Germany significantly influence and hence determine the results.
Table 5
Regressions of cash holdings on language.
The sample consist of listed firms within the EURO-zone during the period of 2010 to 2014. Cash holdings are measured as the ratio of cash and equivalents to total assets. Cash flow is the ratio of operating cash flow to total assets, i.e., ocf/ta. Size is the log of total assets, i.e., log(ta). Liquidity, is the net working capital minus cash to total assets, i.e., (nwc-cash)/ta. Investment size, is the log of R&D expenditures, i.e., log(rd) Strong FTR, is based on the nationality of CFO and CEO and a binary variable that indicates whether their native language grammatically separates future from present, where a value of 1 indicates that a language does and a value of 0 indicates the opposite. All regressions are run using pooled OLS. The expected relationships are noted in the second column. Model (1) to (3) are based regressed employing the full sample. Model (1) is a univariate regression of cash holdings on language, without control variables. Model (2) includes control variables, year dummies, and industry dummies to control for fixed effects. Similarly, model (3) includes dummies and control variable. However, it also includes investment size. Model (4) to (6) have the same specifications as model (1) to (3) with the exception that France and Germany are excluded from the sample. Regression coefficients are presented in the separate columns with the corresponding t-statistics in parentheses. ***, **, * indicates significance levels at the 1%, 5% and the 10% level, respectively.
Full Sample Excluding France & Germany France & Germany
Expected signs (1) (2) (3) (4) (5) (6) (7) (8) (9) StrongFTR - -.019 (-7.05)*** -.015 (-5.27)*** -.008 (-2.24)** -.004 (-1.02) .006 (1.49) .012 (2.42)** -.036 (-8.80)*** -.035 (-8.90)*** -.025 (-4.72)*** Cash Flow + .175 (8.65)*** .162 (5.93)*** .201 (7.47)*** .215 (6.01)*** .176 (5.63)*** .130 (3.19)*** Size - -.005 (-6.55)*** -.017 (-12.09)*** .000 (.019) -.013 (-6.59)*** -.011 (-10.47)*** -.020 (-9.35)*** Liq - -.036 (-4.70)*** -.065 (-4.84)*** -.012 (-1.00) -.023 (-1.30) -.097 (-6.78)*** -.133 (-6.50)*** Leverage - -.009 (-3.70)*** -.013 (-3.39)*** -.004 (-1.09) -.005 (-1.04) -.014 (-4.19)*** -.023 (-3.56)*** Inv Size + .021 (7.32)*** .024 (6.48)*** .016 (3.38)***
Industry and year FE No Yes Yes No Yes Yes No Yes Yes
Obs 3199 3136 1660 1830 1790 822 1369 1346 838
Adj. R2 0.015 0.056 0.139 0 0.034 0.124 0.054 0.154 0.200
Moreover, French and German are two clear opposites in case of FTR, where the first is considered a strong FTR language and the latter a weak FTR language. Besides, German is one of the 6 weak FTR languages, whereas there are 12 strong FTR languages within our sample. From this it can be argued that there is not enough variation within the Euro-zone in terms of FTR, which implies that by removing France and Germany the variation becomes even smaller. Hence, by excluding these values our sample loses strength. To continue on the control variables, we noted that the results are consistent with prior studies as they confirm the expected signs in column 1 table 5. More specifically, corporate cash holdings are negatively related to size, liquidity, and leverage, whilst positively related to both cash flow and investment size.
5. Robustness tests
To confirm the robustness of our findings we perform various additional tests. The results of these tests are summarized in Table 6. To enhance comparability and ensure compactness, we only emphasise upon the coefficients of our main explanatory variable. This idea is derived from Chen et al. (2015). Hence, we only report language in table 6 and do not re-examine the control variables. However, they are still included in all depicted models.1
First of all, by employing the methodology of Chen et al. (2015) we run statistical tests with alternative measures of cash holdings. These alternative measures: the natural logarithm of cash and equivalents, the ratio of cash and equivalents over sales revenue, and the natural logarithm of the cash ratio are derived from Opler et al. (1999), Bates et al. (2009), and Chang and Noorbakhsh (2009). These alternative measures generate similar results compared to our original measure. Both alternatives measures are significantly negative at the one-percent level with coefficients of -.046 and -.116 respectively. Moreover, it reinforces our hypothesis that there is a negative relationship between FTR and cash holdings. Alternatively, if we exclude France and Germany, column 2 table 6, we find a similar positive relationship to earlier in this study. This confirms that France and Germany are overrepresented in this study.
Additionally, we use alternative measures for the level of FTR within a specific language. Hitherto, research on economic behaviour and language is rather thin, however, Chen (2013) provides two alternative measures. First of all, is the sentence ratio, which is a ratio that measures the ‘share of sentences contain a
grammatical future marker’. (Chen, 2013, pp 727) Secondly, is a verb ratio that measures the amount of
verbs that are grammatically future marked and is computed as a ratio to the total number of verbs within an exert of text. The results of these tests can be found in panel B table 6. However, the output of both ratios are insignificant for the full sample, which may have several implications. First of all, it indicates that both
1 For a more detailed description of the models, see the subscript of Table 6
the verb ratio and the sentence ratio are not reliable alternative measures of language. However, the ratios are significant for the smaller subsample excluding France and Germany. Secondly, it implies that language is not a significant determinant of cash holding at the firm level.
Finally, we use a random effects panel regression to address the possibility of omitted variables in our analysis. (Hill, Griffiths, and Lim, 2008) Furthermore, it is important to remember that we are unable to conduct fixed-effect panel regression as our main explanatory variable FTR is time-invariant. The corresponding output is illustrated in panel C table 6. The findings are similar to prior, a negative relationship is found for the entire sample and a positive for the smaller subsample excluding France and Germany. However, the latter is statistically insignificant, whilst the expected negative relationship is statistically significant at the one-percent level.
Table 6
Robustness tests.
The sample consist of listed firms within the EURO-zone during the period of 2010 to 2014. All regressions are run using pooled OLS, unless stated otherwise. All regressions include the following control variables; Cash flow is the ratio of operating cash flow to total assets, i.e., ocf/ta. Size is the log of total assets, i.e., log(ta). Liquidity, is the net working capital minus cash to total assets, i.e., (nwc-cash)/ta. Strong FTR, is based on the nationality of CFO and CEO and a binary variable that indicates whether their native language grammatically separates future from present, where a value of 1 indicates that a language does and a value of 0 indicates the opposite. Additionally, all models, except for the random effect model, control for year and industry fixed effects through industry and year dummies. The Strong FTR coefficients are displayed with t-statistics (in parentheses). ***, **, * indicates significance levels at the 1%, 5% and the 10% level, respectively.
Full sample Excluding France and Germany
FTR FTR
Panel A:
Alternative measures of cash holdings
Log of cash and equivalents -.046(-2.68)*** .050(2.03)** Ratio cash and equivalents over sales Revenue -.016(-1.41) .037(5.36)***
Log of Cash Ratio -.116(-3.13)*** .089(1.67)*
Panel B:
Alternative measures of language
Sentence ratio -.0002(-1.25) .0004(5.49)***
Verb ratio -.0002(-1.29) .0004(5.52)***
Panel C:
Alternative regression specification
Random effects panel -.016(-2.89)*** .003(0.47)
6. Sectoral analysis
Prior in our study we stated that the level of cash holdings between industries may differ for various reasons. Moreover, prior research has mainly focussed on the level of cash holdings on a country level (Opler et al. 1999; Ozkan, A. and Ozkan, N., 2004; Bates et al, 2009; Chen et al., 2015) Hence, in this section we aim to investigate the relationship between language and cash holdings at a sectoral level. Remember, table 3 depicts the summary statistics across the sectors and illustrates a few interesting differences between the sectors. Table 7 provides the regression output of our sectoral analysis. For this analysis we use the same model without investment size as prior in this study with the exception of employing industry dummies. We do not include industry dummies as the observations are already categorized into three separate sectors. However, we do include country dummies to control for other country specific effects, which allows us to use the entire sample including France and Germany Additionally, the reason for excluding investment size is the relatively small sample size of the primary sector.
Table 7
Regression of cash holdings on language per sector.
The sample consist of listed firms within the EURO-zone during the period of 2010 to 2014. Cash holdings are measured as the ratio of cash and equivalents to total assets. Cash flow is the ratio of operating cash flow to total assets, i.e., ocf/ta. Size is the log of total assets, i.e., log(ta). Liquidity, is the net working capital minus cash to total assets, i.e., (nwc-cash)/ta. Strong FTR, is based on the nationality of CFO and CEO and a binary variable that indicates whether their native language grammatically separates future from present, where a value of 1 indicates that a language does and a value of 0 indicates the opposite. All regressions are run using pooled OLS and include both year and country dummies. Regression coefficients are presented in the separate columns, per sector, with the corresponding t-statistics in parentheses. ***, **, * indicates significance levels at the 1%, 5% and the 10% level, respectively.
Primary Manufacturing Secondary
Strong FTR .118 (0.79)*** -.012 (-3.53)*** -.040 (-1.68)* Cash Flow .026 (0.38) .222 (8.98)*** .302 (6.48)*** Size .001 (0.39) -.006 (-6.26)*** -.007 (-4.82)*** Liquidity -.191 (-6.37)*** -.027 (-2.45)** -.044 (-2.62)*** Leverage -.001 (-0.15) -.008 (-2.93)*** -.017 (-3.30)***
Year and country FE Yes Yes Yes
Obs 145 2253 738
Adj. R2 0.367 0.086 0.24
To start with our main variable of interest, strong FTR, the table indicates some interesting results. First of all, there is a contrast between the primary sector and both the manufacturing and the services sector. For the latter two sectors we find a significant negative relationship at the one-percent level and the ten-percent level respectively, whilst the findings for the primary sector suggest there is a significant positive
relationship at the one-percent level between FTR within a language and corporate cash holdings. This may confirm that there are inter-sectoral differences. However, in line with our prior regressions we find the same significant relationships for all control variables across the sectors. In the secondary and tertiary sectors size, liquidity, and leverage have a statistically significant negative effect on cash holdings, and cash flow illustrates a significant positive relationship. For the primary sector, we only find a significant effect for liquidity, whilst the other variables are insignificant. However, we must note that this may be caused by the relatively small sample size.
7. Conclusion
This study is the first to analyse the relationship between language and cash holdings. Thereby, this paper aims to explain why organizations have high levels of cash. Moreover, this paper is one of the first to study the determinants of cash holdings on a sectoral level. Our language hypothesis is based on the work by Chen (2013) who studied the effect language on economic behaviour. Therefore, we hypothesise the level of corporate cash holdings is higher for MNEs with executives from countries with weak FTR languages. More specifically, languages that do not grammatically differentiate between future and present. To test this hypothesis we use a sample of 637 listed companies in the EURO-zone over a period of 5 years from 2010 to 2014. For the full-sample we find a statistically significant negative relationship between FTR and cash holdings. However, when we exclude France and Germany from the sample to ensure they are not overrepresented in the sample and influence the statistical tests, we find contradicting results. These regression indicate there is a significant positive relationship between FTR and corporate cash holdings. Moreover, this distinction is confirmed during our robustness tests. First of all, by employing alternative specifications for cash holdings we find similar results in both samples. However, for the full sample one specification, namely the ratio of cash and equivalents of sales revenue, is insignificant. On the contrary, all alternative measures underline the positive relationship in the sample excluding France and Germany. Moreover, when examining the output for the alternative measures of language no significant relationships are found for the full sample, whilst the positive relationship is confirmed for the smaller sample. An important issue highlighted by Chen (2013) is that it is possible that language does not cause but reflects deeper differences that drive economic behaviour in individuals, such as savings. This might explain the different statistical outcomes. Even though, we can partially confirm our hypothesis as we find that there is a significant relationship between language and cash holdings. Moreover, Dahl (2000) noted that languages are relatively stable over time, e.g. Germanic languages do not grammatically separate future from present since approximately 2,000 years ago. Hence, executives may align financial policies with their own cultural values, which are reflected through their native language. However, to adequately confirm the expected sign of this relationship, either positive or negative, future research is advised.
Besides, this study emphasises on investment size as organizations are becoming increasingly high-tech due to technological advances. (Amess et al., 2015) As a result larger investments are necessary, which increases precautionary cash holdings. Hence, we hypothesized that a positive relationship exists between investment size and corporate cash holdings. By employing a sample of 1660 observations of 332 firms from the period 2010 to 2014, we confirm our hypothesis and find that investment size is significantly positively related to the level of cash holdings. Moreover, by excluding France and Germany from the sample due to their large effect on the sample we find the same result. Therefore, we can accurately conclude that investment size is an important determinant for the level of cash holdings. In combination with the notion that industries are becoming increasingly high-tech (Amess et al., 2015), this study provides statistical evidence for the rise in cash holdings over the past decades. Furthermore, this solidifies the theory of the precautionary motive for holding additional cash.
In sum this study provides additional evidence for the agency theory and the precautionary motive as it indicates that the language spoken by executives, agents, and the investment levels influence corporate decision-making. However, it must be noted that this study has limitations. The main limitation concerns the variance in our explanatory variable FTR within the Euro-zone, which is reflected by the relatively small number of different languages spoken by executives in the Euro-zone. Moreover, CEOs and CFOs may have changed between 2010-2014. However, this would only influence the results if the two following conditions are both met; First, the executive needs to be replaced within a relatively short time-frame. Second and more important, the new executive needs to have a different nationality and thus a different FTR score. However, due to lack of data availability and the relatively short time-span in which this study was conducted it is beyond the scope of this paper to analyse board circulation. Therefore, this study assumes that the executives did not change and hence uses the median FTR score of executives within a certain company. Additionally, this study emphasises on the cash holdings in listed firms, whereas the relationship between cash holdings and language might be different in non-listed firms as these firms generally hold less cash (Von Eije, 2012). More extensive future studies should take into account the circulation of executives. Besides that, a more global sample is advised as this may create more variation in the FTR variable. Another option is the inclusion of dialects as these are often numerous within countries.
9. References
Almeida, H., Campello, M., Weisbach, M.S., 2004. The cash flow sensitivity of cash. Journal of Finance 59, 1777–1804.
Amess, K., Banerji, S., Lampousis, A., 2015. Corporate cash holdings: causes and consequences. International Review of Financial Analysis 42, 421–433.
Bates, T., Kahle, K., Stulz, R., 2009.Why do U.S. firms hold so much more cash than they used to? Journal of Finance 64, 1985–2021.
Bliss, B.A., Cheng, Y., Denis, D.J., 2015. Corporate payout, cash retention, and the supply of credit: Evidence from the 2008-2009 credit crisis. Journal of Financial Economics 115, 521–540.
Brooks, C., 2014. Introductory econometrics for finance. Cambridge University Press, Cambridge.
Brown, J., Petersen, B., 2011. Cash holdings and R&D smoothing. Journal of Corporate Finance 17, 694– 709.
Cooper, M.J., Jackson III, W.E., Patterson, G.A., 2003. Evidence of predictability in the cross-section of bank stock returns. Journal of Banking & Finance 27, 817–850.
Chang, K., Noorbakhsh, A., 2009. Does national culture affect international corporate cash holdings? Journal of Multinational Financial Management 19, 323–342.
Chemmanur, T.J., Fulghieri, P., 1994. Reputation, renegotiation and the choice between bank loans and publicly traded debt. The Review of Financial Studies 7, 475–506.
Chen, K., 2013. The effect of language on economical behaviour: evidence from savings rates, health behaviors and retirement assets. American Economic Review 103, 690–731.
Chen, Y., Dou, P.Y., Rhee, S.G., Truong, C., Veeraragavan, M., 2015. National culture and corporate cash holdings around the world. Journal of Banking and Finance 50, 1–18.
Deloitte, 2014. The cash paradox: How record cash reserves are influencing corporate behaviour. Available at: <https://www2.deloitte.com/content/dam/Deloitte/uk/Documents/corporate-finance/deloitte-uk-cash-paradox-jan-14.pdf> [Accessed at: 26 February 2016].
Deloitte, 2015. The Australian cash paradox. Available at: <http://www2.deloitte.com/au/en/pages/media-releases/articles/australian-cash-paradox-lazy-capital-delivering-lazy-growth-250215.html> [Accessed at: 15 March 2016]
Duchin, R., Ozbas, O., Sensoy, B., 2010. Costly external finance, corporate investment and the subprime mortgage credit crisis. Journal of Financial Economics 97, 418–435.
Han, S., Qiu, J., 2007. Corporate precautionary cash holdings. Journal of Corporate Finance 13, 43–57. Hill, R.C., Griffiths, W.E., Lim, G.C., 2008. Principles of Econometrics. John Wiley & Sons, Hoboken, NJ. Himmelberg, C.P., Petersen, B.C., 1994. R&D and internal finance: a panel study of small firms in high tech industries. Review of Economics and Statistics 76, 38–51.
Hofstede, G., 1980. Culture’s Consequences: International Differences in Work-related Values. Sage Publication, Beverly Hills, CA.
Hofstede, G., 2001. Culture’s consequences: Comparing values, behaviors, institutions, and organizations across nations. Sage Publication, Beverly Hills, CA.
Jaruzelski, B., Schwartz, K., Staack, V., 2015. Innovation’s new world order. Available at: <http://www.strategy-business.com/feature/00370?gko=e606a> [Accessed at: 14 March 2016].
Kim, C., Mauer, D., Sherman, A., 1998. The determinants of corporate liquidity: Theory and evidence. Journal of Financial and Quantitative Analysis 33, 335–359.
Miller, M.H., Orr, D., 1966. A model of the demand for money by firms. Quarterly Journal of Economics 80, 413–435.
Myers, S., Majluf, N., 1984. Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics 13, 187–221.
Newbold, P., Carlson, W., Thorne, B., 2012. Statistics for business and economics: Global edition. Pearson Education Limited, Harlow, UK.
Opler, T., Pinkowitz, L., Stulz, R., Williamson, R., 1999. The determinants and implications of corporate cash holdings. Journal of Financial Economics 52, 3–46.
Ozkan, A., Ozkan, N., 2004. Corporate cash holdings: An empirical investigation of UK companies. Journal of Banking and Finance 28, 2103–2134.
Pinkowitz, L., Williamson, R., 2001. Bank power and cash holdings: Evidence from Japan. Review of Financial Studies 14, 1059–1082.
Ross, S.A., Westerfield, R.W., Jaffe, J., Bradford, J., 2011. Corporate finance: core principles and applications. McGraw-Hill.
Song, K., Lee, Y., 2012. Long-term effects of a financial crisis: Evidence from cash holdings of east Asian firms. Journal of Financial and Quantitative Analysis 47, 617–641.
Stulz, R., Williamson, R., 2003. Culture, openness, and finance. Journal of Financial Economics 70, 313– 349.
Von Eije, J.H., 2012. What causes differences between listed and unlisted firms around the world. Available at SSRN: <http://ssrn.com/abstract=2010053> [Accessed at: 3 March 2016]
