Does international diversification decrease the cash holdings of US firms?

(1)

1

Master Thesis

Does international diversification decrease the cash holdings

of US firms?

Amsterdam Business School

MSc Finance, Quantitative Finance track

Luoran Wu (11252030)

Date: 01-07-2018

(2)

2

Statement of originality

This document is written by Student Luoran Wu, who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

This study examines the impact of international diversification on corporate cash holdings for US firms from 2000 to 2017. I find that corporate cash holdings will decrease as the degree of international diversification enhances, and this adverse effect of international diversification is stronger for only international diversified subsample. Moreover, for the financially constrained firms, the adverse effect of international diversification on corporate cash holding is more pronounced than the financially unconstrained firms. However, the cash holdings of financially unconstrained firms become more sensitive to changes in international diversification, given firm with high product diversification. The results also suggest that for financially constrained firms, international diversification strategy is a good substitute for product diversification strategy.

(4)

4

The determinants of corporate cash holding have been widely discussed in the existing corporate finance literature. The decision regarding how much cash to hold is very crucial for a firm since cash is the most liquidity assets, and it could cover your debt and fulfill your immediate needs. Holding cash could help firms saving its transaction cost to raise funds and avoiding assets liquidation to fulfill the obligation of payment. The cash reserve can also be used to finance the profitable investment opportunities and activities when external finance is not available or too costly. However, according to the trade-off theory, holding cash could bring benefits to a firm, it also generates a cost. The cost including the lower rate of return since cash is liquid assets, and it could generate tax disadvantage as you have the more tax-deductible amount. Therefore, it is an undecisive question that whether the firm should hold more cash and how much should be held.

Since the determinants of corporate cash holdings have been debated in the literature for a long time, some theories have been put forward to support the decision of cash holding. A previous study, Bates et al. (2009), summarizes four motives for firms to hold cash, which are precautionary savings, reduction in transaction cost, agency problem and tax motive. They found the most robust evidence for precautionary demand while no evidence for agency problem or taxes. According to Foley et al. (2007), US multinational firms hold a significant amount of cash because of the tax payment associated with repatriating foreign income, which supports the argument that multinational firms tend to accumulate cash. Moreover, the tax motive of holding cash to repatriate foreign income sheds light on my research question regarding whether the more international diversified firms tend to hold more cash.

International diversification could help firms improve their internal capital market. The work of Duchin (2010) shows that the internal cash flows between divisions could allow firms are becoming less dependent on corporate cash reserves. However, previous studies mostly link diversification with firm performance. The impact of firm diversification on corporate cash holding does not receive much attention. Also, the effect of diversification on cash holding is not very straightforward. It is worth investigating the relationship between the degree of diversification and cash holding. In addition, diversification can be measured in different dimensions. Product diversification and international diversification are different dimensions of

(6)

6

diversification. Product diversification is about firm has a different variety of product lines. For instance, the sales revenue of a shop only selling sunglasses depends on the weather condition. In this case, this shop has no product diversification. However, it could expand their business into selling umbrella as well, such that the revenue the shop is independent of the weather condition to some extent. Having a variety of business line helps a firm diversifying their risk in front of exogenous shocks. The firm could utilize product diversification or expand their business lines to hold less precautionary savings, which is also supported by the existing literature. I would expect that international diversification ought to have a similar effect to corporate cash holdings since the reasoning for having international diversification is similar for product diversification in terms of diversified risk. International diversification could be defined as the expansion of a firm beyond the borders of its home country across different countries and geographic regions (Capar and Kotabe, 2003). Firms have some extent of international diversification having at least two segments operating in different countries, and it could cope with economic fluctuation better because there is less possibility that all countries will be affected by the financial crisis at the same time. Consequently, the whole cash flow of a firm will become smoother if they adopt an international diversification strategy, and the company could have less cash holding in case of lousy scenario.

My empirical study is primarily based on the cash holding behavior of US firms regarding the role of international diversification. The databases used for the study are the COMPUSTAT North America file and Historical segments file. These databases are widely used in the existing literature. However, the sample period used in my study is more recent compared to previous studies. There are two hypotheses needed to be tested for answering the research question that how international diversification determines corporate cash holdings. The first hypothesis is that greater international diversification lowers the corporate cash holdings, and the second hypothesis is that cash holding is more sensitive to international diversification for the financially constrained firms compared to the financially unconstrained firms. According to Duchin (2010), the year fixed effects and the firm fixed effects are controlled in the regressions. In my modified methodology, the two above mentioned hypotheses are tested by using OLS regressions with industry fixed effects. The second hypothesis is tested by categorizing the full sample into two subsamples, which are the financially

(7)

7

constrained and the financially unconstrained firms, respectively. Besides that, the variable of interest is the international diversification instead of the product diversification. Apart from the firm characteristics controlled in Duchin’s (2010) regression model, cash flow ratio and R&D expenses are taken into account in my study as well. To further investigate the interactive effects of international and product diversification, the dummy variable of high product diversification has been controlled in one of the regression models. By interacting the product diversification dummy with the proxies for international diversification, the exact effect of international diversification on corporate cash holdings, given firms already have high product diversification, can be investigated.

In sum, a new angle, firm’s international diversification, will be studied for determining the corporate cash holdings. By controlling both international and product diversification in the empirical model, I can find their segregate effects on corporate cash holdings. Moreover, financial constraints can considerably affect the pattern of cash holding across firms. Specifically, if firms are financially constrained, they would have more incentives to hold precautionary cash to avoid forgoing profitable investment opportunities. While for financially unconstrained firms, they have more accesses to external financing, therefore, fewer incentives to hold cash. It is expected that differential effect of international diversification can be found in these two groups of firms. By dividing the full sample into two subsamples, namely financial constrained and unconstrained, based on the corporate payout ratio, whether the sensitivity of cash holding to international diversification differs in these two groups will be examined. By implementing the methodology explained above, I find evidence that foreign sales to total sales ratio (FSTS) should be a necessary control variable for the primary regression model instead of being an indicator for international diversification. Because positive coefficient has been found for FSTS, which contrast with the theoretical expectation. However, I found supportive evidence that more international diversification significantly lowers corporate cash holding when using foreign assets to total assets ratio (FATA) or geographic segment ratio (HHI) as the proxy, keeping other factors constant. Moreover, by using FATA and HHI together for international diversification, I found the overall effect of international diversification on corporate cash holding, which is almost twice as the effect when the single indicator is used. Only looking at the effect captured by FATA, the effect of more international diversification

(8)

8

on cash holding is roughly the same regardless of financially constrained firm and financially unconstrained firm. Financially constrained firms tend to hold less cash than financially unconstrained firms when considering the effect captured by both FATA and HHI, which is consistent with the second hypothesis. However, given firms are highly product-diversified, enhancement in international diversification continues lowering the cash holding of the financially constrained firms. While for the financially unconstrained firms, their cash holding increases for the increases in international diversification if they have high product diversification.

The contribution of my thesis will be the following. Firstly, my study fills the gap of international diversification in corporate cash holding literature. Most of the diversification researches focus on product diversification of firms. The unique point of my empirical study is that I will focus on how international diversification affects the corporate cash holdings. Since there is no literature exactly studying the international diversification and corporate cash holdings, it is imperative to perform an empirical study regarding this topic to fill the gap in the literature. Besides that, product diversification is controlled throughout all the regressions in my study, which allows a comparison to be made between the effect of product and international diversification. Secondly, it is the first study using multiple indicators for international diversification to conduct the empirical analysis. There are several measures used in previous literature to capture the international diversification. Using multiple indicators within the model for international diversification can capture the effect on corporate cash holding comprehensively.

My empirical study proceeds as follows. In section 2, the most relevant literature regarding the research topics such as cash holdings, international diversification, and product diversification are reviewed. Besides that, the development of hypothesis is explained in this section as well. In section 3, the data sources and the descriptive statistics are presented. In section 4, the methodology used for the study is explained. In section 5, the empirical results for all relevant regressions are presented. Lastly, section 6 is the robustness checks for empirical findings. In section 7, the conclusion will be drawn given the results from the previous section.

(9)

9

2. Literature review and hypothesis development

2.1 Determinants of corporate cash holdings

The determinants and implications of cash holdings had received considerable attention in the existing literature. Why firms hold cash? There are four motives for firms holding cash based on the review of Bates et al. (2009). First, the benefit for firms holding cash is to save transaction cost in terms of liquidating noncash financial assets to make payment. Keynes (1936) describes it as transaction cost motive of corporate cash holdings. Second, firms could use cash to finance its business activities, and sound investment opportunities when facing lousy state and access to the capital market is very costly. This motive is also recognized as the precautionary motive of holding cash. Third, the tax motive, as Foley et al. (2007) find that US corporations have a higher level of cash holdings due to high tax expenses generated from repatriating foreign income. Their finding suggests that multinational firms have a high probability of holding more cash. Fourth, agency motive. Specifically, as Dittmar et al. (2003) find that corporations which located in countries with weak corporate governance tend to hold more cash. Dittmar and Mahrt-Smith (2007) show that governance has a critical influence on the value of cash holdings, well governance firms almost double the value of cash compare to poorly governance firms. In addition, Harford et al. (2008) provide evidence of weakly controlled managers tend to spend cash quickly than stronger governance manager given same high level of cash holdings. The stronger governance firms tend to increase dividends and commit higher payout ratios in the future. However, the weak governance firms tend to repurchase rather than a commitment on dividends. Their study suggests that weakly controlled managers are more likely to generate high corporate cash holdings.

Using sample of publicly traded US firms from 1971 to 1994, Opler et al. (1999) show that firms with substantial growth opportunities and riskier cash flows have relatively high cash holdings, and their results also show that large firms and firms with better credit ratings hold relatively less cash due to secure access to the capital market. These results are in line with the precautionary motive view that firms tend to hold more cash to keep investing when cash flow is low, and the access of capital market is expensive.

(10)

10

2.2 International diversification

International diversification is a strategy of firms, which diversify their business geographically. It implies that firm adopting international diversification strategy will set up foreign subsidiaries in order to expand the sales of goods or services overseas. Hence, international diversification is a synonym for geographic diversification, internationalization or multi-nationality. International diversification could help firms to reduce their return fluctuations by spreading their investment risk internationally (Kim et al.,1993). Less fluctuation in return suggests firm could hold less cash since stable return will weaken the precautionary motive for cash holdings. Lu and Beamish (2004) found that the extent of geographic diversification is negatively associated with firm performance when the level of internationalization is high and low. While at a moderate level of internationalization, greater geographic diversity generates higher firm performance. Hence, I expect that better performance firms tend to hold less cash because of stable cash flow or easy to excess external capital market.

Foley et al. (2007) give another perspective on determinants of corporate cash holdings. Their study indicates that tax motive also plays a significant role on corporate cash holdings. They found that US multinational firms that are facing more substantial tax expenses by repatriating earnings hold more cash. Base on their findings, it is plausible that multinational firms will hold more cash because of tax motivation. Bodnar et al. (1997) proposed that it is more costly and challenging monitoring of managerial decision making in a sophisticated and international diversified firm. This is consistent with the conclusions of Burgman (1996) such that international-diversified firms have higher agency costs compare to non-diversified firms. In addition, Burgman (1996) also found that multinational firms do not have lower earnings volatility due to international diversification strategy. Based on his results, Multinational Corporations will have more incentive to lower their cash holding so as to eliminate the higher agency cost issue. However, his results also suggest that international diversification could not help the firms to lower cash holding through lower earnings volatility. The effect of international diversification on cash holdings of firms is ambiguous given the work of Burgman (1996).

Contrast with previous literature, Reeb et al. (1998) found evidence that international-diversified firm may increase their systematic risk owing to an increase in cash flow

(11)

11

volatility from international-diversified strategy. The results of Duru and Reeb (2002) shows that multinationals in the US have approximately 65% greater variation in earnings volatility compare to domestic firms, which also suggest multinationals will hold more cash due to earnings volatility.

2.3 Diversification and corporate cash holdings

Divisional diversification or product diversification is another dimension of diversification strategy of a firm. Each segment of the diversified firm has independent investment opportunity hence it has an internal capital market in the diversified firm (Khanna and Tice, 2001). The independent investment opportunity implies that international-diversified firm needs less cash on hand to meet their investment need since they could finance by cash flow from another foreign subsidiary. According to Pecking order theory (Myers, 1984), firms are more willing to use the internal market to finance their investment instead of using external financing, which also supports international-diversified firm might hold less cash than non-diversified firm.

Duchin (2010) found out multidivisional firms hold less cash than specialized firms because they are diversified in their investment opportunities. In his study, he argues that diversified firms tend to have more smooth investment opportunity and cash flows since investment opportunities and returns from different divisions are not perfectly correlated. Therefore, multidivisional firms will have less precautionary cash holdings. Indeed, his empirical results are in line with his argument. Another key finding of him is that for financially constrained firms lower cash holdings mainly results from diversification. This finding suggests that diversification strategy could be a substitute for the precautionary motive of holding cash.

Furthermore, Tong (2011) shows that firm diversification is associated with a lower value of cash in both financially constrained and unconstrained firms. Therefore, I expect that international diversification will increase the value of cash holdings in financially constrained firms. Pinkowitz et al. (2012) providing a detailed analysis regarding the cash holdings of US multinational firms. The analysis indicates that the high cash holdings of multinational firms are not due to the tax expenses of foreign earnings repatriations. Furthermore, their empirical results also do not show any evidence of poor governance lead abnormal cash holdings of US multinational firms. International diversification could improve the internal capital market. Duchin (2010)

(12)

12

provides evidence that diversification could improve the internal capital market by transferring the excess cash of divisions that do not have investment needs at a particular time period to the divisions that have investment needs. This internal cash flows between divisions could allow firms are becoming less dependent on corporate cash reserves. Hence it will help the firms reducing the precautionary demand for cash. For international-diversified firms, different subsidiaries also diversify their investment opportunities internationally. I expect they could also transfer excess cash from abundant cash subsidiaries to subsidiaries that having investment needs.

2.4 Hypothesis development

According to the literature review described above, I formed two hypotheses for my empirical study. The rationale of each hypothesis is listed below.

Hypothesis 1: International diversification will decrease the corporate cash holdings.

Base on previous literature, international-diversified firms will tend to hold less cash because of less precautionary demand for cash holdings. Specifically, international-diversified firms tend to have more smooth investment opportunities and cash flows because the cash flow and investment opportunities are less correlated between different foreign subsidiaries. Meanwhile, the internal capital market is also enhanced by international diversification strategy since cash could transfer across subsidiaries. It will flow from subsidiaries without current investment opportunities to subsidiaries with better investment opportunities. A fundamental motive of precautionary cash holdings for a corporation is fear to forgo good investment opportunities when facing adverse shocks. When the internal flow of cash is possible, international-diversified firms will reduce their cash holdings. Note that there are very crucial assumptions behind. Namely, all investment opportunities in a firm could not entirely be financed externally, and the external funds are expensive compared to internal funds. Since if the firm could finance their valuable project internally, it is unnecessary to borrow money externally, and if we live in a perfect financial market world, there is costless to borrow money outside.

Hypothesis 2: The cash holdings are more sensitive to the changes in international

diversification for financially constrained firms than financially unconstrained firms. Financially unconstrained firms have more accesses to external financing compare to financially constrained firms. Therefore, I expect that financially unconstrained firms,

(13)

13

their cash holdings might less affect by international diversification than financially constrained firms. Financially constrained firms to hold more precautionary cash to prevent themselves from forgoing good investment opportunities when experiences adverse cash flow shocks and external financing for them is even more costly compared to financially unconstrained firms such that they have more incentive to hold precautionary cash. In opportunity cost perspective, the cost of holding cash is the firm might be forced to forgo profitable project today, and it is even more costly for financially constrained firms compared to the financially unconstrained firm; they face more difficult to find valuable investment opportunities. Therefore, the opportunity cost of holding cash is the cost of giving up the valuable investment today. The benefit of holding cash is the firm remains the ability to finance the foreseeable investment opportunity in the future. In addition, the opportunity cost of not holding cash for the financially constrained firm is relative high, since they are unlikely to get external finance if they are facing adverse cash flow shocks compare to the financially unconstrained firm.

To some extent, International diversification could alleviate the motivation of holding more precautionary saving, and it could lower the opportunity cost of not holding cash to the financially constrained firm. In sum, if the financially constrained firm is international diversified, the effect of cash holding reduce is expected stronger than financially unconstrained firms.

3. Sample selection and data description

3.1 Sample selection

The studied samples in this study are primarily from the COMPUSTAT’s North America file and Historical Segment file compiled by Standard and Poor’s. Specifically, the studied sample including long-lived assets, total assets, current assets, research and development expenses, net sales, segment identifier, standard classification code, capital expenditures, common/ordinary equity, cash and short-term investment, common shares outstanding, current liabilities, debt in current liabilities, total long-term debt, depreciation and amortization, total dividends, income before extraordinary items, purchase of common and preferred stock, deferred taxes, total market value and annual close price for almost all of the available American firms from the COMPUSTAT database. The data are firm annually and cover the fiscal year 2000 to

(14)

14

2017. The COMPUSTAT measure the financial data of firms in millions of USD and different geographic and business segments data for individual firms are included in the segment dataset. Majority of firm-level data will be gathered from COMPUSTAT Fundamental Annual File to calculate the dependent and most of the explanatory variables, and the segment-level data will be gathered from COMPUSTAT’s Historical Segment File. The data from COMPUSTAT’s Historical Segment are mainly collected for calculating the key independent variables (international diversification and product diversification variable). The two datasets will be merged together into the final sample set for this paper.

Furthermore, utility firms with Standard Industrial Code between 4900 and 6999 are excluded from the study as well as financial firms with Standard Industrial Code from 6000 to 6999. Besides that, firms with missing data on cash holdings are also excluded from the sample. The primary reason for excluding utility and financial firms is to be consistent with previous research. Excluding financial firms from the study due to the fact that these firms are subject to the statutory capital requirement and they hold inventories of marketable securities that are included in cash (Foley et al., 2007). The reason for excluding utility firms is because their cash holdings are under regulatory supervision. I also followed the research of Almeida et al. (2004), firms with missing cash holdings data and the firm market capitalization lower than 10 million USD are eliminated from the study. Finally, merging the two COMPUSTAT’s file together, the final sample of this study include 12,536 firm-year observations from 2000 to 2017.

3.2 Data description

The dependent variable, namely, the cash holdings of the firm will be measured by cash and short-term investment divided by assets. The most important explanatory variables in the study are the proxies for international diversification and the proxy for product diversification. Firstly, it is worth to mention the most commonly used proxy for international diversification. The Foreign sales to total sales ratio (FSTS) are widely used a proxy for international diversification in the literature (e.g., Duru and Reeb (2002); Denis et al. (2002)), and FSTS ratio will scale from 0 to 1 indicating that the degree of international diversification. In other words, the larger the FSTS, the higher the degree of international diversification. However, the FSTS ratio is not ideal to measure the international diversification because it mixes up the international export sales with foreign subsidiaries sales. We could not distinguish whether the sales amount

(15)

15

comes from export or foreign subsidiaries base on the Foreign Sales to Total Sales ratio. As suggested by Reeb et al. (1998), using Foreign Assets to Total Assets ratio (FATA) could mitigate the issue of mixing up export and subsidiaries sales, since international trade number could not affect this measure. The FATA is similar financial ratio compare to FSTS, it will scale from 0 to 1, and 1 representing the highest degree of international diversification. The larger the FATA means, the more assets a company has overseas, and it also reflects that more international diversified a company is. Since there is previous literature discussed that the limitation of their study measures the international diversification by a single indicator such as FSTS ratio, it is desirable to have more indicators for international diversification to reflect the effect on cash holdings better. An alternative way to measure international diversification is using an HHI measure, which is quite similar to the measurement of product diversification used in this study. This indicator of international diversification is geographic segments sales ratio using an HHI calculation method. Followed the methodology of Kang et al. (2017) the specific calculation of this measurement is following,

International diversification = 1 − ∑(𝑆𝑎𝑙𝑒𝑠𝑗 ∑𝑆𝑎𝑙𝑒𝑠𝑗)

2_,

Where 𝑆𝑎𝑙𝑒𝑠_𝑗 indicate the geographic segment sales from segment j, and the firm and year subscripts are omitted. This proxy of international diversification also ranges from 0 to 1, the higher the index the higher the international diversification. The geographic segments sales ratio (HHI) together with foreign assets to total assets ratio should capture the effect of international diversification better than a single indicator.

The product diversification variable will be calculated by a Herfindahl-Hirschman measure (product diversification = 1 − ∑ 𝑃_𝑖2, where 𝑃_𝑖 is the proportion of a segment sales on total sales) following prior study (Tallman and Li, 1996). Note that the firm and year subscripts are also omitted in here. The product diversification variable will be put into the regression equation in order to better capture the effect of international diversification on corporate cash holdings.

Furthermore, other control variables which determine the corporate cash holdings will be included in the regression. All those control variables are scaled by the firm size. Cash flow affects the corporate cash holding considerably because the cash flow of a firm through the fund availability influences the feasibility of valuable investments.

(16)

16

The investment opportunity itself also play a role in determining the corporate cash holdings, and the proxy of investment opportunity is Tobin’s Q in this study. In addition, a larger firm might need the more substantial amount of cash holdings to compare to a small firm. Therefore, the firm size also matters for cash holdings. If it is costly to raise fund from the financial market because of asymmetric information, the firm has to contract investment opportunity given shortfall of cash flow. It is expected that firm with high research and development (R&D) expense tend to hold more liquid assets such as cash to lower their cost of forgone valuable investment opportunity when facing financial distress. As a consequence, I expect the more R&D expense a firm has, the more cash it would hold. The other explanatory variables such as book leverage, payout ratio, and capital expenditure will also affect the cash holding level. Therefore, they are considered when studying the impact of international diversification on corporate cash holdings. The calculation of most of the control variables will follow the work of Duchin (2010). On top of his work, the R&D expenses, foreign sales to total sales ratio, foreign assets to total assets ratio, geographic segments ration are included in the regression of this paper. Another difference is I use difference measure of product diversification, and he did not study international diversification. In appendix A, the detailed definition and calculation of used variables in the regression are stated.

Panel A in Table 1 presents the descriptive statistic of all variables in the studied sample. Including the information regarding the number of observation, mean, median, standard deviation, minimum and maximum of variables. This table reveals the primary variables in this study, cash holdings, has substantial variation among different firms, it varies from 0 to 97%. The mean firm has cash holdings 18.5% of its total assets, and the deviation of cash holdings is 17.7%, and the median (12.6%) suggests that the distribution of cash holdings is right-skewed. The foreign sales to total sales ratio (FSTS) and the foreign assets to total assets ratio (FATA) are in line with the expectation that they all scaled from 0 to 1 to indicate the degree of firm diversification. Again, the FSTS is not the ideal proxy for international diversification, but it could still reflect part of the diversification degree to some extent. If the FSTS equals to zero, then there is no international diversification for a firm. the geographic segment sales HHI indicate the Herfindahl measure of geographic segments sales, and it could be a good proxy for international diversification.

(17)

17

Variable Obs Mean Median Std. Dev. Min Max

Dependent variable:

Cash Holdings (cash/assets) 12,536 0.1848 0.1263 0.1765 0 0.9669

Independent variables:

Foreign sales/total sales (FSTS) 12,536 0.4565 0.4225 0.2957 0 1

Foreign assets/total assets (FATA) 12,536 0.3919 0.3055 0.3272 0 1

Geographic segment sales (HHI) 12,536 0.4872 0.5151 0.2273 0 0.8560

Product Diversification 12,536 0.3386 0.3930 0.2975 0 0.8509 Tobin's Q 11,074 1.7521 1.5379 0.8083 0.4171 7.5176 Book Leverage 12,483 0.2049 0.1779 0.2177 0 10.4800 Payout/assets 11,829 0.0434 0.0176 0.0694 0 0.4476 CAPEX/assets 12,494 0.0422 0.0302 0.0410 0 0.5091 NWC (exluding cash)/assets 12,384 0.0772 0.0748 0.1583 -1.7061 0.5030 R&D/assets 9,067 0.0536 0.0336 0.0584 0 0.6325

Firm Size (ln(assets)) 12,536 7.1637 7.0657 1.9122 1.2890 13.1124

Cash flow ratio 12,517 0.1113 0.0977 0.0627 0.0050 0.4187

Cash

holdings FSTS FATA HHI PD Tobin's Q

Book leverage Cash holdings 1 FSTS 0.1691 1 FATA -0.0589 0.5094 1 HHI 0.0357 0.6236 0.2242 1 PD -0.2963 -0.0197 0.0175 0.0967 1 Tobin's Q 0.3612 0.0121 -0.0562 -0.0044 -0.2201 1 Book leverage -0.4355 -0.0478 0.0572 0.0122 0.1704 -0.1616 1 Payout ratio 0.1007 -0.0171 -0.0622 0.0085 -0.052 0.3173 0.0071 CAPEX -0.1512 0.0293 0.059 -0.0123 -0.0519 0.0526 0.0297 NWC -0.2922 -0.1224 -0.0342 -0.0555 0.0211 -0.1672 -0.1272 R&D 0.491 0.1661 -0.0702 0.1229 -0.2054 0.3042 -0.2831 Firm size -0.2945 0.1242 0.1268 0.1955 0.3328 -0.0945 0.3488 Cash flow 0.171 0.0244 -0.008 -0.0375 -0.1439 0.4799 -0.1323

Table 1. Descriptive statistics

Panel A: Summary statistics

This table reports the summary statistics for the whole studies sample, which consists of nonfinancial and nonutility firms from 2000 to 2017. The firms are all with non-missing data on cash holdings and business or geographic segments sales. The market capitalization of each individual firm is above 10 million USD. All financial variables reported blew are scaled by firms to make them comparable, and they all winsorized at 1% level on either tail. The FATA is the proxy for international diversification, and HHI is an alternative measure of international diversification.

Panel B: Correlation matrix

This table represents the partial correlation matrix of variables used in this empirical study. Cash holding is the dependent variable, and the others are explanatory variables. FATA is a proxy for international diversification, and the variable HHI is an alternative proxy for international diversification measured by HHI method. FSTS is an important control variable, and it could help the FATA better reflect the relationship between international diversification and cash holdings.

(18)

18

This variable and product diversification variable are also varied from 0 to 1 to reflect the degree of diversification. The rest of the explanatory variables are falling into a reasonable range when compared with previous studies. Moreover, all financial variables are winsorized at 1% level to alleviate the effects of outliers.

Panel B is the partial correlation matrix for the variables used in the empirical study. The cash holding, FATA, and HHI are the critical variables of interest. Based on the correlation coefficient of HHI on cash holdings, the relationship between the degree of firms’ international diversification and cash holding is positive. In other words, it is plausible that firms which are more internationally diversified are associated with holding more cash.

From the perspective of Foreign assets to total assets ratio (FATA), the cash holdings of firms decrease in foreign assets to total assets ratio. Since both of those two variables capture the international diversification of a firm, we could not learn the relationship between international diversification and cash holdings from this correlation table directly. In the meanwhile, the correlation coefficient of HHI on FATA is 0.2242 which suggest that there is no multicollinearity issue. Besides that, larger firms with higher capital expenditures, higher net working capital, more leverage-takings, holding more foreign assets and more product diversified tend to have lower cash holdings. While for firms with higher foreign sales, more international diversified, more investment opportunities, a higher payout ratio, more investments in R&D and more cash flow, their cash holdings tend to be higher. Table 1 gives some insights regarding the expected effect of each variable on cash holding, which will be further studied in the following section.

4. Methodology

In this section, the methodology of investigating the effect of international diversification on corporate cash holdings is explained. The empirical study is based on a sample with observations at the firm-year level, and fixed effects panel regression model will be implemented. The sample period is from the year 2000 to 2017. By following the methodology of Duchin (2010), two regression models are used for my empirical study.

(19)

19

𝐶𝑎𝑠ℎ ℎ𝑜𝑙𝑑𝑖𝑛𝑔𝑠_𝑖𝑡 = 𝛽₁𝑖𝑛𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙_𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑓𝑖𝑐𝑎𝑡𝑖𝑜𝑛_𝑖𝑡+ 𝛿_𝑗𝑐𝑜𝑛𝑡𝑟𝑜𝑙_{𝑖𝑡,𝑗}+ 𝜆_𝑘+ 𝑢_𝑖𝑡 The full sample is a panel data, and each observation corresponds to the corporate cash holding for firm i in year t. There are three indicators for international diversification, namely, FSTS, FATA, and HHI, respectively. These three proxies will be used separately as the proxy for international diversification such that the effect of international diversification on corporate cash holding can be captured.

In addition, a regression model with multiple indicators for international diversification will be implemented as follows:

𝐶𝑎𝑠ℎ ℎ𝑜𝑙𝑑𝑖𝑛𝑔𝑠_𝑖𝑡 = 𝛽₁𝐹𝐴𝑇𝐴_𝑖𝑡+ 𝛽₂𝐻𝐻𝐼_𝑖𝑡+ 𝛿_𝑗𝑐𝑜𝑛𝑡𝑟𝑜𝑙_{𝑖𝑡,𝑗}+ 𝜆_𝑘+ 𝑢_𝑖𝑡

, where FATA stands for foreign assets to total assets ratio for firm i in year t. HHI stands for geographic segments sales ratio. The variable FATA together with variable HHI will capture the effect of international diversification from sales and assets, respectively. The control variables are indexed with letter j. 𝜆_𝑘 is the industry fixed effects with k represents each industry. 𝑢_𝑖𝑡 stands for the error term, which includes all the other items determining corporate cash holdings. To investigate the relationship between the international diversification and cash holding, it is assumed that all the explanatory variables outlined are independent of the error term.

Generally speaking, the international diversification is only measured by a single indicator in previous literature. The measure of foreign sales to total sales (FSTS) is commonly chosen for representing the degree of international diversity. However, it is not ideal to use only one single indicator, and it could not capture all the international diversity of a firm. As can be seen from the correlation table in section 3, the foreign assets to total assets ratio (FATA), the foreign sales to total sales ratio (FSTS), the geographic segment sales ratio (HHI) all have a correlation between each other (around 0.20). In this case, it is not feasible to drop either one of those three variables otherwise the rest of the variable will have a correlation with error term then generate endogeneity problem. The FSTS, FATA, and HHI all capture the degree of international diversity of a firm from different perspectives. The ratio of foreign sales to total assets (FSTS) is controlled because the definition of this measure mixes up the international trade with the sales from foreign subsidiaries. If the foreign sales ratio is removed from the regression given the fact that the geographic segment sales ratio (HHI) is included in the regression, the regression will face endogeneity problem. Therefore, FSTS, as a

(20)

20

useful control variable for corporate cash holding, is kept in the regressions to avoiding omitted variable bias. The FATA could be a proxy for international diversification from assets holding perspective, which measures the proportion of foreign assets to total firm assets. This measurement could reflect how much assets a firm have overseas, where foreign assets must come from the foreign subsidiaries. Therefore, the higher the FATA ratio, the higher the degree of international diversity. Neither one of the variable itself could explain the whole degree of international diversity, but all them together could help to capture the effect of international diversity of a firm on their cash holdings better.

Besides that, some other factors, which have explanatory power to corporate cash holdings, have been controlled in most of the regressions. Those control variables are commonly used in previous studies, which are Tobin’s Q, book leverage ratio, payout ratio, capital expenditures, net-working-capital excluding cash, R&D expenses, firm size and cash flow. Tobin’s Q is a widely-used proxy for corporate investment opportunities. Holding more cash enables firms financing more valuable investment instead of forgoing them because of fund shortage. Thus, it is expected that firms with abundant investment opportunities are more inclined to hold more cash. Firm leverage ratio can also explain the corporate cash holdings. For firms with a high leverage ratio, it is necessary that firms have sufficient funds available to pay the fixed interest payment. If the obligations of interest payments cannot be met, then firms will have to default against the debt, which destroys the reputation and affect the stability of the firm. Therefore, for highly leveraged firms, it is expected that their cash holding will be higher on average to secure the debt payments. Moreover, it is expected that firms with higher payout ratio tend to hold less cash because there is a need to distribute profits to shareholders such that cash at hand will decrease. Furthermore, it is expected that firms with higher capital expenditures and net-working-capital will hold less cash. Also, multinationals are on average more substantial in size compared to domestic firms. To make the analysis more comparable across firms, firm size is controlled in the empirical model. For all the regressions, industry fixed effects are controlled to capture the factors which differ across industries, which could alleviate the omitted bias problem as well.

The empirical study will be implemented in the following steps. Firstly, regressions in Table 3 and Table 4 test for the first hypothesis, which states that international

(21)

21

diversification will decrease corporate cash holding. By performing three different regressions in full samples and only diversified firms sample separately, three different proxies for international diversification will be implemented. To be specific, the first regression of full samples or only diversified firms sample will use FSTS as the proxy for international diversification. Using FSTS aim to compare the results with the previous study as the FSTS is widely used for international diversification in the existing literature. Since the shortcoming of using FSTS, the foreign sales to total sales ratio (FATA) is used in the second regression instead of FSTS for both full firm sample and only diversified firm sample. In theory, the geographic segment sales ratio (HHI) can also be the proxy for international diversification, since it reflects how much sales foreign subsidiaries generated. The HHI measure the degree of international diversification from a different perspective. The third regression will only use HHI as the proxy for international diversification to examine the effect of international diversification on corporate cash holdings. Further, the three measures of international diversification will concatenate together to study the effect of more intense international diversification on corporate cash holdings. Intuitively, FSTS includes not only foreign sales but also foreign exporting and HHI reflect the concentration degree of foreign sales to total sales of a firm. Hence, FSTS could be a good control in the regression to help other two measurements of international diversification better capture the effect of international diversification on corporate cash holdings.

Besides that, two subsamples, financially constrained and financially unconstrained firms, will be constructed from the full sample. The full sample is divided into ten deciles by using payout ratio as a criterion. Based on Fazzari et al. (1988), the financially constrained firms are more inclined to have lower payout ratio, and firms with higher payout ratio are financially unconstrained. Therefore, the highest three and the lowest three will form the financially unconstrained group and constrained group, respectively. The aim of further divided full sample into two subsamples is to test the second hypothesis that whether within financially constrained firms corporate cash holding is more sensitive to the changes in the degree of international diversification. In addition, by interacting highly-product-diversified dummy with international diversification, I can study whether being a highly-product-diversified firm would weaken the need for being international diversified under each subsample.

(22)

22

5. Empirical results

The analysis is primarily based on the results presented in four tables. Table 2 reports the cash holdings for each measurement of international diversification and each group by using the difference-in-means method. Table 3 presents the OLS regression results regarding the effect of international diversification on cash holding by merely using one indicator for international diversification in each regression. After that, in Table 4, multiple indicators are used for international diversification in each regression model to capture the effect comprehensively. Lastly, Table 5 reports the regression results looking at the differential effect of international diversification on corporate cash holdings across financially constrained and unconstrained firms.

5.1 The patterns of cash holdings across samples

Table 2 consists of three panels. The statistics of each panel are based on a different sample, which gives an overview of the mean cash holdings across subgroups of firms. In all three panels, three different proxies, namely, FSTS, FATA, and HHI, are used for international diversification.

Panel A is based on the full sample, where ‘High’ and ‘Low’ international diversified groups are constructed by using different proxies as criteria. Based on each proxy, the whole sample is divided into ten deciles. “High” group includes the observations with the corresponding international diversification measure in the top three deciles. “Low” group includes the observations with the corresponding international diversification measure in the lowest three deciles. The column data for “High” and “Low” groups are the corresponding mean cash holdings by using three different proxies for international diversification. To test whether the average cash holdings for the "High" and "Low" groups are significantly different, I use the difference-in-means method. The null hypothesis of the t-test is the average cash holdings across subgroups are statistically significantly equal. The corresponding t-statistics shows that the null hypothesis has been rejected at 1% significance level for every measure used for international diversification, implying that the average cash holdings for the most and the least international diversified firms are different based on the full sample.

The statistics of Panel B is merely based on the observations in the least international diversified group defined by three different measures. The least international diversified group are further categorized into the financially constrained and

(23)

23 Panel A: All firms

High

Variable High Low minus Low t-Statistics

Foreign sales to total sales ratio (FSTS) 0.2284 0.1446 0.0838 20.7407 Foreign assets to total assets ratio (FATA) 0.1849 0.1994 -0.0145 -3.4545 Geographic segements sales ratio (HHI) 0.2059 0.1584 0.0475 11.8364

Panel B: Low international diversified firm Variable FC firm FU firm FC firm minus FU firm t-Statistics

Foreign sales to total sales ratio (FSTS) 0.1370 0.1695 -0.0324 -5.1729 Foreign assets to total assets ratio (FATA) 0.1831 0.2317 -0.0486 -6.8669 Geographic segements sales ratio (HHI) 0.1570 0.1763 -0.0192 -2.8739

Panel C: High international diversified firm Variable FC firm FU firm FC firm minus FU firm t-Statistics

Foreign sales to total sales ratio (FSTS) 0.2323 0.2479 -0.0156 -2.2587 Foreign assets to total assets ratio (FATA) 0.1881 0.2014 -0.0133 -2.0422 Geographic segements sales ratio (HHI) 0.2039 0.2336 -0.0297 -4.6083

unconstrained firms by using their payout ratio as a criterion. By performing a t-test for the average cash holdings for the financially constrained and unconstrained groups, I could compare whether the average cash holding is significantly different across groups. Again, the t-statistics implies that the null hypothesis of equal average cash holdings across groups is rejected at 1% significance level. In other words, given firms being in the least international diversified group, the pattern of cash holdings is significantly different when they are financially constrained or unconstrained. Also, the statistics show that the financially constrained firms tend to hold less cash compared to the financially unconstrained firms, for all three different proxies.

Table 2. The patterns of cash holdings across samples

This table consists of three panels. The statistics of each panel is based on a different sample. Each panel reports the patterns of cash holdings across groups for different measures. There are three different proxies for international diversification, namely, FSTS, FATA, and HHI. In Panel A, the whole sample is categorized into two groups by using different proxies. ‘High’ and ‘Low’ represents the subsamples of the most and the least diversified firms. In Panel B, the statistics are based on the subsample of the least diversified firms defined by three different proxies. The least diversified firms further divided into ‘FC’ and ‘FU’ groups, representing the groups of the financially constrained and unconstrained firms. Similarly, Panel C is based on the subsample of the most internationally diversified firms, which is further categorized as ‘FC’ and ‘FU’ groups. In each panel, whether cash holdings are significantly different across subgroups is checked by using the difference-in-means method.

(24)

24

Similarly, the statistics of Panel C is based on the observations in the most international diversified group defined by three different measures. Based on the test results, I also find that given firms being in the most international diversified group, their average cash holdings are significantly different when they are financially constrained and unconstrained. The cash holdings of financially constrained firms are still lower than the cash holdings of financially unconstrained firms, given they are all highly international diversified.

In sum, the results in Penal A provide evidence that pattern of cash holdings are significantly different for the most and least international diversified firms. Thus, it is worth studying what the exact effect of international diversification on cash holding is in the following empirical studies. In Panel B and C, I find that for both “High” and “Low” international diversified subsample, the average cash holdings for firms being in the financially constrained and unconstrained subgroup are significantly different. Therefore, it is worth taking the financial constraints of firms into account when studying the effect of international diversification on corporate cash holdings.

5.2 The single indicator for international diversification

The regressions in Table 3 are aiming to study the exact effect of international diversification on corporate cash holdings. The relevant regression models are performed for two samples, namely, the full sample and the sample of international-diversified firms respectively. For each sample, the basic regression model is performed with using different international diversification indicator. All three measures for international diversification scale from 0 to 1, reflecting the degree of international diversification accordingly. The reason for doing similar regressions on the international-diversified firms is because some firms have a pretty low level of international diversification or even they are not international diversified. Therefore, it would be better to exclude those firms from the full sample to eliminate the noisy effects of those firms. Such that, only the firms at least have some degree of international diversification can be used for the empirical study. According to the definition, firms having more than two different foreign segments are qualified for being in the most international diversified group. Briefly, regressions based only on the internationally diversified group could help to find how the changes in the degree of international diversification affect the corporate cash holdings, given that firms are already multinational firms. Except for the key variable of interest, some relevant control

(25)

25

All firms Only international diversified firms

Variables (1) (2) (3) (4) (5) (6) FSTS 0.0457*** 0.0902*** (6.34) (3.76) FATA -0.00554 -0.0386*** (-1.06) (-3.78) HHI 0.00105 -0.0962*** (0.11) (-5.12) PD -0.0558*** -0.0578*** -0.0577*** -0.0116 -0.0131 -0.0140 (-8.80) (-9.12) (-9.13) (-1.03) (-1.18) (-1.26) Tobin’s Q 0.0429*** 0.0421*** 0.0422*** 0.0537*** 0.0525*** 0.0531*** (13.90) (13.68) (13.72) (10.70) (10.56) (10.78) Book leverage -0.297*** -0.303*** -0.303*** -0.248*** -0.247*** -0.245*** (-20.84) (-21.16) (-21.08) (-9.89) (-9.95) (-9.79) Payout ratio 0.00612 -0.00703 -0.00451 0.00685 -0.0285 0.00697 (0.21) (-0.24) (-0.15) (0.13) (-0.53) (0.13) CAPEX -0.737*** -0.734*** -0.735*** -0.723*** -0.713*** -0.743*** (-15.20) (-14.89) (-14.90) (-9.79) (-9.67) (-10.07) NWC (excl. cash) -0.320*** -0.331*** -0.331*** -0.265*** -0.280*** -0.262*** (-15.40) (-15.94) (-15.91) (-6.09) (-6.23) (-6.02) R&D 0.319*** 0.327*** 0.329*** 0.158** 0.130* 0.194*** (5.59) (5.77) (5.77) (2.18) (1.77) (2.71) Firm size -0.0146*** -0.0131*** -0.0133*** -0.0221*** -0.0217*** -0.0197*** (-11.43) (-10.54) (-10.29) (-9.98) (-9.81) (-8.75) Cash flow ratio 0.0108 0.0188 0.0179 0.0338 0.0506 0.0366 (0.25) (0.43) (0.41) (0.47) (0.70) (0.51) Observations 7,408 7,408 7,408 2,569 2,569 2,569 R-squared 0.549 0.546 0.546 0.528 0.529 0.532

Industry FE Yes Yes Yes Yes Yes Yes

Table 3. The single indicator for international diversification

This table reports the OLS regression results for the effect of international diversification on cash holding by using three different proxies. All the dependent variables are corporate cash holdings, which is measured by the sum of cash and short-term investment divided by total book assets. The first three regressions are based on the sample of full firms. Regression (1) uses foreign sales to total sales ratio (FSTS) as the proxy for international diversification. In regression (2), the proxy for international diversification changes into foreign assets to total assets ratio (FATA). In regression (3), another measure, geographical segment sales ratio (HHI), is used as a proxy for international diversification. Regression (4), (5) and (6) reports the results of the same regression models, except that the sample used is merely the international-diversified firms. Firms having at least two foreign subsidiaries are qualified for being in the group. ‘PD’ represents product diversification, and ‘NWC’ represents net working capital excluding cash. Industry fixed effects are controlled in each regression. Robust t-statistics are reported in parentheses. All the standard errors are clustered at the firm level. Significance at 10%, 5% and 1% level are represented by *, ** and ***, respectively.

(26)

26

variables are included in all regression models. Industry fixed effects are controlled as well to capture the characteristics varying across industries.

FSTS is a widely used proxy for international diversification. Therefore the regression results reported in column (1) and (4) can be used to make a comparison with the findings in the previous literature. According to the results, the effect of international diversification on corporate cash holdings, using FSTS as the proxy, is significantly positive. Also, the magnitude of the effect found within the sample of firms with international diversification is twice as much as the effect found in the full sample regression. It is proven that excluding those irrelevant firms in the empirical study enhances the significance of the results. Since the scale of the FSTS is from zero to one, it is more reasonable to assume a 0.1 increase in the scale to interpret the economic meaning of the coefficients. The coefficient of 0.0457 implies that the corporate cash holdings will increase by 0.00457 if FSTS increases by 0.1. Since corporate cash holding is measured by the ratio of the sum of cash and short-term investment to the book assets, an increase of 0.00457 in the ratio can be substantial when firms are relatively large. The positive relationship also implies that the higher the proportion of foreign sales in total sales, the higher the corporate cash holdings.

Although FSTS is widely used in the literature, it is not an ideal proxy for international diversification for the following two reasons. From one perspective, international diversification and product diversification are merely two different dimensions for the study of corporate diversification. Previous literature found negative relationships between product diversification and corporate cash holdings. Therefore the similar effect of international diversification on the corporate cash holdings is expected. However, the opposite effect for international diversification has been found in my empirical study by using FSTS as a proxy. Therefore, the accuracy of using FSTS as an indicator for international diversification is suspicious. From another perspective, the definition of FSTS mixes up the international trade and foreign sales. It is possible that the proportion of international trade dominates compared to the foreign sales portion, such that FSTS could not accurately capture the relationship between international diversification and corporate cash holding.

Additionally, another measure, geographic segments sales ratio (HHI), is used as the third proxy for international diversification. The geographic segments sales ratio is

(27)

27

calculated by using an HHI measurement. This measure implies the concentration level of sales coming from the different geographic region. In my empirical study different geographic region represents different foreign subsidiaries. According to the empirical results in column (3) and (6), the relationship between international diversification and cash holding remains negative as the findings of FATA regressions. However, using HHI as an indicator for international diversification magnifies the effect significantly. A higher HHI implies that firms are more geographically diversified, meaning that sales are coming from more geographic segments. Based on the sample of firms has some degree of international diversification, the coefficient of -0.096 suggests that firms with more geographical diversification hold less cash on average.

Except for the variable of interest, some control variables are included in each regression as well. I find the similar adverse effect of product diversification on corporate cash holding, although it seems like product diversification plays a less important role within those international-diversified firms. Also, I find expected effect of Tobin’s Q, where a higher Tobin’s Q is associated with more cash holdings. As mentioned earlier, firms having more investment opportunities tend to hold more cash to avoid forgoing profitable investment opportunities. Besides that, corporate capital expenditure and net-working-capital are associated with lower cash holdings. Moreover, larger firms tend to hold less cash, which might be because of their easier access to external financing than smaller firms. After controlling industry fixed effects, the explanatory power of each regression model is strong.

In sum, by using three different measures for international diversification, different effects have been found. Based on the results, FSTS might be a good control variable instead of being the indicator for international diversification in the regression model. The definitions of FATA and HHI can more closely capture the notion of international diversification, and they can be further used as the indicators for international diversification. Moreover, because FATA and HHI measure international diversification from different dimensions, it would be better to include both measures in the regression model to capture the effect of international diversification comprehensively.

(28)

28

5.3 Multiple indicators for international diversification

In Table 4, the empirical study is still based on two different samples. What differs from the regressions in the previous table is the inclusion of all three measures for each regression. However, FSTS is not considered as one of the proxies for international diversification. Instead, it is used as a useful control variable for explaining the corporate cash holdings. Empirical results prove that FSTS can explain a large part of corporate cash holdings, therefore controlling it can significantly improve the explanatory power of the model and alleviate the omitted variable bias. On the other hand, FATA and HHI are used together to capture the effect of international diversification.

Such that, FSTS can help these two indicators of international diversification to capture the effect better. Moreover, for each sample, two regressions are performed, where the only difference is the inclusion of year fixed effects. Adding year fixed effects is aiming to capture the factors that vary across different years.

Based on the results, FSTS has a significant positive effect on corporate cash holdings, which is consistent with the previous finding. The effects are magnified compared to results reported in Table 3. The results also show that both FATA and HHI have an adverse and significant effect on corporate cash holdings at 1% significance level, although their effects are more pronounced in the sample of the international-diversified firms. In other words, by using these two indicators together for international diversification, the coefficients imply that more international diversification lowers the corporate cash holdings. This finding also provides supportive evidence for the first hypothesis of my empirical study. To interpret the total effect of international diversification on corporate cash holding, I sum up the coefficients of FATA and HHI. For example, in column (3), the magnitude of the effect in total is -0.1648 (-0.0676-0.0972), implying that the corporate cash holdings ratio will decrease by 1.648% if both indicators increase by 0.1 in total. As mentioned earlier, small changes in the cash holdings ratio can reflect a substantial change in the amount of cash holding, especially for those larger firms with the more substantial amount of book assets.

(29)

29

All firms Only international diversified

firms Variables (1) (2) (3) (4) FSTS 0.0860*** 0.0858*** 0.111*** 0.117*** (8.58) (8.50) (4.26) (4.43) FATA -0.0348*** -0.0352*** -0.0676*** -0.0667*** (-5.47) (-5.53) (-6.06) (-5.98) HHI -0.0484*** -0.0490*** -0.0972*** -0.0971*** (-4.30) (-4.35) (-5.07) (-5.07) PD -0.0542*** -0.0542*** -0.00482 -0.00383 (-8.56) (-8.54) (-0.43) (-0.34) Tobin’s Q 0.0430*** 0.0450*** 0.0534*** 0.0549*** (14.06) (14.24) (11.03) (10.83) Book leverage -0.294*** -0.290*** -0.235*** -0.233*** (-20.75) (-20.28) (-9.53) (-9.35) Payout ratio 0.000925 0.000555 -0.000585 -0.00186 (0.03) (0.02) (-0.01) (-0.03) CAPEX -0.734*** -0.731*** -0.721*** -0.742*** (-15.40) (-14.97) (-10.08) (-10.06) NWC (excl. cash) -0.316*** -0.314*** -0.263*** -0.263*** (-15.32) (-15.35) (-6.09) (-6.22) R&D 0.309*** 0.305*** 0.133* 0.124* (5.45) (5.37) (1.84) (1.71) Firm size -0.0137*** -0.0135*** -0.0197*** -0.0191*** (-10.63) (-10.50) (-8.93) (-8.53)

Cash flow ratio 0.00602 -0.00470 0.0296 0.0214

(0.14) (-0.11) (0.43) (0.30)

Observations 7,408 7,408 2,569 2,569

R-squared 0.553 0.554 0.543 0.545

Industry FE Yes Yes Yes Yes

Year FE Yes Yes

Table 4. Multiple indicators for international diversification

This table reports the effect of all measurements of international diversification on cash holdings. Regression (1) and (2) are based on the full sample, and (3) and (4) are based on only internationally diversified firms. All three measures together in each regression capture an overall effect of international diversification. Industry fixed effects are added in each regression. The additional year fixed effect is added for regression (2) and (4). The variable PD represents for product diversification, and variable NWC represents net working capital excluding cash. Control variables are included in all regressions. Robust t-statistics are reported in parentheses. Significance at 10%, 5% and 1% level are represented by *, ** and ***, respectively.

Does international diversification decrease the cash holdings of US firms?

Master Thesis