The effect of foreign exchange exposure on firms' cash holdings

THE EFFECT OF FOREIGN

EXCHANGE EXPOSURE ON FIRMS’

CASH HOLDINGS

YUANJUE WANG

University of Amsterdam, Amsterdam Business School MSc Finance + Quantitative Finance

Master Thesis 2018.06

1 / 35 Statement of Originality

This document is written by Yuanjue Wang 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.

2 / 35 Part I. Introduction

With the process of globalization, there is an increasing number of firms which operate overseas and are exposed to foreign currency exchange risk due to their international transactions and competition. After the collapse of the Bretton Woods system, which was characterized by widespread restrictions on international capital mobility (Lane et al.,2005), in the 1970s, the fluctuation of major currencies has become increasingly intensive as the exchange rate between US dollars and other currencies has become flexible. In open economies, an increasing number of firms are found to be exposed to the ups and downs of trade-weighted currency index (Jong et al.,2006). Therefore, currency risk has played an increasingly important role in the daily risk management, especially for nonfinancial firms.

On the other hand, according to Bates, Kahle and Stulz (2009), the average cash holding level for firms which are located in the United States has experienced a huge rise from 1980 to 2006, which is mainly driven by precautionary motive. However, their research only focusses on precautionary motive and agency problems. In addition, there are more evidence to show that firms with more investment opportunities and higher cash flow risk, in other words, are financially constrained, tend to have a higher cash ratio (Opleretal,1999). Overall, it is almost convinced that there is a much higher possibility for financially constrained firms that they may have a higher cash ratio.

Therefore, in order to fund themselves for future potential investment, firms tend to hold cash with exposure to foreign exchange fluctuation. When there is an unexpected and unfavorable movement in the exchange rate, firms’ ability to finance their future investment opportunities may be challenged as firm value when it is dominated in foreign currencies will follow the changes in the exchange rate. As a result, faced with the risk that the income of international transactions in foreign currency fluctuates, firms will increase their cash holdings to hedge against such risks.

There has not been a paper which focuses on the effect of foreign exchange exposure on firms’ cash ratio. As Bartram (2008) mentioned in his paper, such an

3 / 35

exposure has an insignificant influence on total cash flow but the operating cash flow is exposed to foreign exchange risk. The insignificance is due to the operations taken by the managers from the nonfinancial firms aiming to reduce the exposure to a level low enough. Additionally, foreign exchange derivatives are also used by managers to reduce foreign exchange exposure as well and are proven to be effective in the short run by Huffman and Makar (2004). However, existed research does not concentrate on the question whether cash holding rather than cash flow is exposed to foreign currency risk.

The thesis will contribute to the study on the cash holding level of nonfinancial firms in the United States. As a result, this thesis can show that how much the firm is going to save from its annual cash flow in order to face the potential foreign exchange risk so that it can still finance itself in the future possible investment opportunities. In addition, in the paper written by Bartram (2008), the extent of risk exposure is measured by the relative change in exchange rate while in this thesis, it will be determined by the regression results of firms’ return on cumulative percentage change in trade-weighted US dollar currency exchange rate index, which is conducted by Huffman and Makar (2004) and Jong et.al (2006).

In this thesis, OLS regressions will be applied to study the effect of foreign exchange exposure, which is measured by the sensitivity of firm’s monthly or daily stock returns to changes in the exchange rate while controlling for market’s return, on nonfinancial firms’ cash holdings. Observations from COMPUSTAT database excluding those financial firms and utilities whose cash levels are set by regulations and those firms who do not operate internationally will be taken into consideration. Consequently, the sign and the significance of the coefficient for the exposure will be viewed as the response of the central research question of the thesis: how does the foreign exchange exposure influence firms’ cash holdings? Is there a difference in the influence between financially constrained and unconstrained firms? To solve the second question, all the observations will be grouped into financially constrained firms and unconstrained ones based on three different financial constraint criteria: dividend payout ratio, asset tangibility and Kaplan-Zingales Index. Consequently, it is confirmed

4 / 35

that the degree of foreign exchange exposure affects nonfinancial firms’ cash holdings positively. And for the second question, it seems that financially constrained firms react more greatly faced to the fluctuations in exchange rate index.

In the next part, relevant previous literature will be reviewed, as well as the hypothesis made based on the existed research. To the next, empirical methodology and the sample is discussed in Part III, following by the summary statistics and a correlation matrix of all the variables related in the regression model. Additionally, how the variables are defined from the original data will be presented in the Appendix after the part of bibliography. Most importantly, the trend of fluctuation of cash holding of nonfinancial firms and the results came from the regression equation are studied in Part IV while Part V provides the robustness check for the regression results offered in the previous part. In the end, conclusions will be presented in the last part, which is Part VI.

Part II. Literature Review

In order to study the effect on the cash holdings, it is necessary to study what determines how much cash a firm holds from studies conducted by other researchers in the past. Previous theory has shown that there are four motives that can be used to explain the change in cash ratio: transaction, precautionary, tax and agency. Firstly, the transaction motive is mainly driven by economies of scale because transaction costs are usually incurred when a firm converts a non-cash financial asset into cash and uses it for payments(Mulligan,1997). Secondly, precautionary cash holding is to deal with potential investment opportunities when it is costly for a firm to obtain funds externally. An increase in the instability in cash flows will lead to an increase in cash holdings for financially constrained firms while it does not have an impact on other firms which are not constrained(Han&Qiu,2007). This point of view is doubted by Riddick and Whited (2009) since it lacks the adjustments for measurement errors in q. One more recent study conducted by Acharya, Almeida and Campello (2007) indicates that when a firm’s income is weakly correlated to its investment opportunities, it prefers internal

5 / 35

funds rather than external financing. Thirdly, when a firm faces a higher repatriation foreign taxes, it is likely to have a higher cash-to-asset ratio. This motive is more deeply studied by Foley et.al (2007). They have found that firms that have a large amount of tax expenses when repatriating earnings are more possible to raise their cash holdings. Furthermore, low tax rates will lead to high repatriation tax expenses, which will also end up with high cash ratios. Lastly, as Dittmar and Mahrt-Smith (2007) have studied, poorly governed firms will value cash less than its real value, resulting in a large cash holding, which is based on the conclusion made by Jensen (1996) that greater agency problems will lead to more cash holdings. Additionally, firms which are located in a country where shareholders’ rights are not protected adequately tend to hold cash twice as much as those in countries with little agency problems (Dittmar, Mahrt-Smith&Servaes,2003). According to the research conducted by Bates, Kahle and Stulz (2009), the cash to asset ratio is mainly based on the precautionary motive and they have also found no evidence which shows that agency conflicts play a significant role in the fluctuation of cash holdings.

In addition, Almeida, Campello and Weisbach (2004) have studied questions about the cash flow sensitivity of cash for all manufacturing firms from the year of 1971 to 2000.They have found that for financially constrained firms, an increase in cash flows will result in a rise in cash holding as well because they have difficulty in access to external capital market when they applied payout policy, asset size, bond ratings or commercial paper ratings as the criteria to partition the sample into unconstrained and constrained subsamples. As a result, such firms will prefer to save more cash from their cash flows every year to meet the financial requirements of possible investment opportunities with positive NPV. However, when the index measure derived from Kaplan and Zingales (1997) is adopted as the classification scheme, both constrained and unconstrained firms have a negative cash flow sensitivity of cash. In addition, when there is a negative macroeconomic shock, an increasing propensity for cash holding of constrained firms has been proved. There is another method to separate financially firms from unconstrained firms: asset tangibility. For firms with a high ratio of tangible

6 / 35

asset to total asset, they are considered as having more pledgeable assets which can support more borrowing as collateral. In other words, there is a small probability for such firms to have difficulty in access to external financial market as the large amount of tangible assets has a negative impact on their risk of financial distress so that they are more likely to be financially unconstrained (Almeida& Campello,2007).

On the other hand, to investigate the effect of foreign exchange on firms’ cash holdings, it is necessary to study the foreign exchange risk, its determinants and its impact on firms. The foreign exchange risk is defined as the probability of loss due to the unfavorable movements of exchange rate. Such risk is a result of multinational firms’ foreign transactions or bonds (Sudacevschi, 2017). For the foreign exchange exposure, it is defined as the amounts of foreign currencies which represent the sensitivity of the home currency value of any physical or financial asset to fluctuations in the future domestic purchasing powers of these foreign currencies in the future (Adler& Dumas,1984). In the paper written by Faff and Marshall (2005), multinational firms that operate more internationally will have a larger magnitude of foreign exchange exposure and there lies difference in the determination of foreign exchange exposure. Even the contribution of international trades to total transactions has a positive relation with the magnitude of foreign exchange exposure for firms in the United Kingdom and a negative relation for multinational firms in Asia-Pacific area, it does not play a role for firms in the United States.

Under the influence of foreign exchange risk, the volatility of cash flow in international operations will rise. As a result, the firm value will reduce if the firm is financially constrained, in other words, has difficulty in raising external funds (Froot et.al,1993).However, it is surprising that a negligible impact of movements of exchange rate on firms’ values across the world has been found by Griffin and Stulz(2001) .Their findings mainly concentrate on the valuation based on stock market rather than firms’ economic activities so that the neglectable influence of foreign currency risk is due to the fact that stock market may ignore the importance of exchange rate risk. Furthermore, the contradistinction may be explained by the condition that foreign currency shocks

7 / 35

do influence firms’ margin profit of export greatly but are not considered to be economically important enough to affect shareholders’ wealth. Lastly, this can also because firms’ managers take actions to minimize the effect of foreign exchange exposure and adjust the firms’ activities to face the fluctuations in exchange rates.

Issuing foreign debt can be viewed as one of the ways of hedging. It is usually assumed that firms who issue foreign currency-dominated debts are naturally hedged (Geczy et.al ,1997). Aside from the usage of foreign debt, derivative hedge is also viewed as one of the method to hedge against movement of exchange rate as well. As Geczy et.al (1997) have also mentioned in their studies, firms with more future growth opportunities or more difficulties in obtaining external funds are more willing to use currency derivatives when they are exposed to foreign exchange risk more deeply. Besides, Allayannis and Ofek (2001) have pointed out that how much the firms are exposed to foreign sales and international trades urges firms to hedge and has a decisive influence on the amount to hedge. While firms are exposed to the foreign exchange risk caused by foreign currency-based activities or competition, the effect on comprehensive performance is mitigated by the actions taken by managers, leading to an economically and statistically insignificant influence on the total cash flow, which is magnificent on operating cash flow (Bartram,2008). It can be concluded from their study that such risk is hedged by managers of nonfinancial firms. However, in the paper written by Jong, Ligterink and Macrae (2006), they find insignificant effects of derivative hedging for Dutch firms, which means that off-balance sheet derivative usage does not reduce their foreign exchange exposure while on-balance sheet hedging plays an important role. Brown (2001) has brought out a research concerning about managing foreign exchange risk with derivatives more in detail. He focused on the management in one single large multinational cooperation, a method viewed as more precise understanding of the process of foreign exchange risk management. From his point of view, exchange rate volatility and exposure volatility have played an important role in the determination of the firm’s hedge portfolio as well as managerial views and recent hedging history. The firm takes options as its first choice in the structuring of its hedge portfolio but for

8 / 35

currencies whose options are less viable, the firm taken as the sample in the paper will make a decision not to hedge other than hedging with forward contracts.

Based on the previous literature, the following hypothesis will be made:

1. Firms which are more exposed to foreign exchange risk will have a higher cash ratio.

2. The cash holding for financially constrained firms will be more sensitive to the fluctuation of foreign exchange exposure.

Part III. Methodology and Data

In this thesis, all COMPUSTAT firms that are located in the United States from the year of 2000 to 2015 except those financial and utility firms (whose SIC code ranges from 4900 to 4999 and 6000 to 6999) will be excluded because their cash holdings depend on capital requirements rather than their demand function depending on their characteristics. Besides, observations with negative asset and sales will be excluded. Last but not least, firms whose foreign income to net income is less than 20% will be removed as well because they are regarded as not exposed to foreign currency risk (Donnelly and Sheehy, 1996).

The level of foreign exchange exposure will be obtained from following equation following Huffman and Makar (2004):

𝑅𝑅𝑖𝑖 = 𝛽𝛽0+ 𝛽𝛽1𝑖𝑖𝐶𝐶𝐶𝐶𝑅𝑅 + 𝛽𝛽2𝑖𝑖𝑅𝑅𝑚𝑚+ 𝜀𝜀𝑖𝑖 (1)

where the exposure is the absolute value of the coefficient of the cumulative percentage change in trade-weighted US dollar currency exchange rate index.CER refer to the cumulative percentage change in trade-weighted US dollar currency exchange rate index (obtained from FRED). Foreign currency exposure will be measured over different time horizons, which are, in this paper, daily and monthly. Therefore, there are two measures for foreign exchange exposure derived from daily returns and monthly returns as well.

9 / 35

foreign exchange exposure can be generated from the absolute value of the beta of the change in currency exchange rate index. Just like what Huffman and Makar (2004) did in their research, the variable which show how much the firm is exposed to foreign currency exchange risk will be added into the demand function as one of the determinants of nonfinancial firms’ cash holdings.

The determinants can be divided into two categories: the firm-level ones and the economic-level one. The firm-specific variables are listed as follows: market-to-book ratio (which will be expressed as Q in the regression model), cash flow to assets, net working capital (nwc) to assets, asset tangibility, leverage, R&D to sales, foreign income to net income, hedge performance, capital expenditures and acquisitions to assets. If there are missing valued in the research and development expenditure, those will be replaced with the value of zero. Based on previous research, the risk caused by foreign exchange exposure may be hedged by derivatives. Therefore, a measure of hedge performance which is derived from the gain or loss from derivatives should be included into the regression model. How the control variables are determined with the original data collected from COMPUTAT is shown in the Appendix. On the other hand, the macroeconomic-level determinant is log GDP, which can be collected from FRED. Additionally, lagged variables will be applied in order to avoid the potential simultaneous causality problem so that the data in t-1 will be used to evaluate the cash holding at time t as it takes time for firms to react according to the macroeconomic surroundings.

The cash holding of a firm in a certain year will be defined cash and marketable securities to book value of total asset (Almeida, Campello&Weisbach,2004; Bates, Kahle&Stulz,2009). This measure will be taken into consideration in this thesis and will be expressed as cash ratio.

The baseline model is constructed as the following equation:

CashHoldings = 𝛽𝛽0+ ∑10𝑖𝑖=1𝛽𝛽𝑖𝑖 × 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐶𝐶ℎ𝑎𝑎𝐹𝐹𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝐹𝐹𝐹𝐹𝑎𝑎𝑎𝑎𝐹𝐹𝑎𝑎𝑎𝑎𝑖𝑖 + 𝛽𝛽11× 𝐺𝐺𝐺𝐺𝐺𝐺+ 𝛽𝛽12𝐹𝐹𝐹𝐹𝐶𝐶𝑖𝑖+

ε (2) where FXE refers to the level of foreign exchange exposure which is calculated from

10 / 35

the equation (1) and firm characteristics are listed in the Appendix. GDP refers to log GDP one year before.

In order to test the second hypothesis, there are three financial constraints criteria based on literature published by Fazzari, Glenn and Bruce (1988), Lamont, Polk and Saa-Requejo (2001), Almeida and Campello (2006), which are dividend payout ratio, asset tangibility and KZ index, which is based on results in Kaplan and Zingales (1997). The criterion of dividend payout ratio was firstly pointed out by Fazzari, Hubbard and Petersen (1988) because firms that pay dividends instead of saving cash are likely not financially constrained. All the nonfinancial firms will be ranked based on their dividend payout ratio and asset tangibility respectively and those in the top two quantiles of the total distribution will be assigned as financially constrained. In other words, firms with less dividend payout or less tangible asset which can be seen to be collateral for external funds are convinced to be financially constrained as there will be a higher probability for such firms that they are more difficult to raise external financing in the capital market .Lastly, an index of the probability that a firm faces financial constraints can also be constructed according to the following equation based on the results in Kaplan and Zingales(1997) and Lamont, Polk and Saa-Requejo (2001): 𝐾𝐾𝐾𝐾 𝐹𝐹𝑖𝑖𝑖𝑖𝑎𝑎𝑖𝑖 = −1.002 × 𝐶𝐶𝑎𝑎𝑎𝑎ℎ𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 + 0.283 × 𝑄𝑄 + 3.139 × 𝐿𝐿𝑎𝑎𝐿𝐿𝑎𝑎𝐹𝐹𝑎𝑎𝐿𝐿𝑎𝑎 − 39.368 × 𝐺𝐺𝐹𝐹𝐿𝐿𝐹𝐹𝑖𝑖𝑎𝑎𝑖𝑖𝑖𝑖 − 1.315 × 𝐶𝐶𝑎𝑎𝑎𝑎ℎ𝐻𝐻𝐹𝐹𝐹𝐹𝑖𝑖𝐹𝐹𝑖𝑖𝐿𝐿𝑎𝑎 (3)

Overall, as dividend payout ratio and asset tangibility have been regarded as the criteria of financial constraints, they will be excluded from the OLS regression model (equation (2)). Therefore, the regression model for each subsample will be the same as the equation (4) listed in the following:

CashHoldings = 𝛽𝛽0+ ∑8𝑖𝑖=1𝛽𝛽𝑖𝑖 × 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐶𝐶ℎ𝑎𝑎𝐹𝐹𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝐹𝐹𝐹𝐹𝑎𝑎𝑎𝑎𝐹𝐹𝑎𝑎𝑎𝑎𝑖𝑖 + 𝛽𝛽9× 𝐺𝐺𝐺𝐺𝐺𝐺+ 𝛽𝛽10𝐹𝐹𝐹𝐹𝐶𝐶 + ε

(4 )

Table I. Descriptive Statistics

This table displays summary statistics for OLS estimators of the regression model for cash holdings (equation (2)). The data (except FXE) are from Compustat database and Federal Reserve Economic Database while the variable FXE is based on the regression results in equation (1). The sample firms exclude financial (SICs 4900 to 4999) and utility (SICs 6000 to 6999) firms and the sample period is from 2000 to 2015. Firms with negative asset or sales and those firms with foreign income to net income less than 20% are removed as well.

11 / 35

Variable Observation Mean Std.

Dev. Min Max Cash Ratio 9,058 0.171697 0.161627 0 0.972853 Asset Tangibility 9,058 0.790964 0.184264 0.095428 1 Foreign Income 9,058 1.755252 39.17196 0.2 2723.875 Acquisition 9,058 0.030303 0.06675 -0.1665 0.748506 Dividend 9,058 0.192193 6.866555 0 0.753733 RD sales 9,058 0.389066 22.79194 0 2156 Leverage 9,058 0.209202 0.184274 0 3.229208 NWC to Asset 9,058 0.117579 0.168448 -1.80411 0.848777 Q 9,058 1.913443 1.234432 0.286417 21.11384

Cash Flow to Asset 9,058 0.073038 0.114039 -3.8377 0.481946

Firm Size 9,058 7.395291 1.897198 1.433178 13.58957 Hedge Performance 9,058 -0.01226 1.090141 -97.8 15.52174 Capex 9,058 0.043884 0.044684 0.000203 0.717533 𝑭𝑭𝑭𝑭𝑭𝑭𝟏𝟏 9,058 0.042102 0.054366 0.002089 0.535252 𝑭𝑭𝑭𝑭𝑭𝑭𝟑𝟑𝟑𝟑 9,058 2.516927 2.911769 0 97.3558 𝒓𝒓𝟏𝟏𝟏𝟏 19,294,549 0.018623 0.031916 0.01 10.94729 𝒓𝒓𝟏𝟏𝟏𝟏 19,294,549 0.000166 0.012175 -0.09188 0.11523 𝑪𝑪𝑭𝑭𝑪𝑪𝟏𝟏 19,294,549 -0.00047 0.002937 -0.02271 0.017491 𝒓𝒓𝟑𝟑𝟑𝟑𝟏𝟏 108,696 0.012117 0.287915 -0.99981 0.99 𝒓𝒓𝟑𝟑𝟑𝟑𝟏𝟏 108,696 0.002979 0.046462 -0.18611 0.112566 𝑪𝑪𝑭𝑭𝑪𝑪𝟑𝟑𝟑𝟑 108,696 -0.0002 0.016099 -0.0389 0.066877

Table I represents the descriptive statistics of the variables mentioned in the model (equation (1) and (2)). The variables which represent the degree of foreign exchange exposure are determined as the absolute value of the coefficient of the cumulative percentage change in trade-weighted US dollar currency exchange rate index in equation (1). The last six rows show the summary statistics of the factors which will define the extent of foreign exchange exposure. As 𝐹𝐹𝐹𝐹𝐶𝐶₃₀ is the yearly foreign exchange exposure derived from monthly firm return, market return and percentage change in trade-weighted US dollar currency exchange rate index, the number of observations will be 12 times of that of variables in equation (2). It can be inferred from Table I that there are firms whose monthly return is not correlated to the change in exchange rates when controlling for market return. The most exposed firm’s return will fluctuate by around 0.74% with one percentage change in exchange rate. Besides, FXE_𝟏𝟏 refers to the annual foreign exchange exposure for each firm which is based on daily stock return, market return and trade-weighted US dollar currency exchange rate index.

12 / 35

Different from the one according to monthly observations, it seems that FXE_𝟏𝟏 has a comparatively smaller range.

While for the firm-specific control variables included in equation (2) and (3) which are derived from items in firms’ annual reports, firms which hold only tangible assets and no intangible assets such as patents, copyrights, goodwill or brand recognition can be discovered from the asset tangibility equal to one. Acquisition to asset shows the cash outflows associated with acquisitions and the negative value shows that there are firms gaining from cash inflows related to acquisition activities. The minimum values of zero in dividend payout ratio, R&D to sales and leverage represent that there exist observations for nonfinancial firms which pay no dividend out or do not have leverage or do not spend on research and development campaign. However, there are also firms which pays a huge amount for research and development, resulting in the maximum value of RD sales of 2156.Turning to the ratio of net working capital to total asset, it shows asset that can be viewed as substitution for cash. However, the negative net working capital to asset indicates that firms have liabilities for cash instead of cash. The variable of q (market-to-book ratio) measures investment opportunities so that a larger market-to-book ratio leads to a higher probability of future positive NPV projects, meaning that value created from investment exceeds the firm’s initial cost of purchasing capital. As firm size is defined as the natural logarithm of total asset, firms whose asset is less than e (about 2.71828) million U.S. dollars will obtain a firm size less than zero. Inferring from the summary of hedge performance, there are both firms that gain from hedging activities with derivatives and firms that performance poorly in derivative hedge as well.

Figure I is a scatter plot of the log of GDP of the United States during the sample period, from the beginning of 21st century to the year of 2015.It can be easily discovered that the gross domestic product in the United States follows an increasing trend generally except in the several years after the financial crisis in the year of 2008.Understandablly, suffered from the recession due to the subprime mortgage crisis, the index has increased a decrease in the crisis years. In the later years, under the

13 / 35

influence of recovery policies taken by the government of the United States, the index shown in Figure I still keeps rising in the 2010s.

Figure I. Log GDP by Year

This figure represents the log of annual GDP of the United States from the year of 2000 to 2015.

From Table II, there will not be problem of multicollinearity between independent variables as there is no correlation larger than 0.5. Most importantly which can be inferred from Table III.The correlation of 0.0013 means that actions taken by managers to hedge the risks caused by changes in the exchange rate which are measured by the gain or loss originated from foreign exchange derivatives are not negatively correlated with the extent of firms’ foreign exposure, which disagrees with the existed findings (Allayannis and Ofek, 2001; Elliott et al., 2003Huffman and Makar,2004; Batram,2008).

If the two hypothesizes made in the last part are true, the regression model using all the nonfinancial firms will end up with a positive relationship between

1 0 0 0 0 1 2 0 0 0 1 4 0 0 0 1 6 0 0 0 1 8 0 0 0 G D P A 2000 2002 2004 2006 2008 2010 2012 2014 year

14 / 35 Table II. Correlation Matrix

This table represents the correlation matrix between all of the estimators in the regression model for cash holdings (equation (2)). The sample firms exclude financial (SICs 4900 to 4999) and utility (SICs 6000 to 6999) firms and the sample period is 2000 to 2015. Firms with negative asset or sales and those firms with foreign income to net income less than 20% are removed as well.

asset tangible foreign income acquisition to asset dividend RD sales leverage nwc to asset q cash flow to asset firm size hedge performance

capex FXE FXEdaily GDP

asset tangible 1 foreign income 0.0067 1 acquisition to asset -0.3431 -0.0067 1 dividend -0.0091 0.004 -0.0006 1 RD sales 0.0033 -0.0001 -0.006 -0.0004 1 leverage -0.1946 0.0183 0.1062 0.0095 -0.0152 1 nwc to asset 0.2736 -0.0107 -0.0768 -0.0054 -0.0187 -0.2472 1 q 0.0024 -0.0169 0.0045 0.0058 0.0028 -0.058 -0.0605 1 cash flow to asset -0.0702 -0.0105 0.0439 -0.0127 -0.0968 -0.1065 0.0905 0.0923 1 firm size -0.1954 -0.0064 0.0534 0.0036 -0.0281 0.2066 -0.2166 0.0257 0.2629 1 hedge performance 0.0136 -0.6067 0.0015 -0.1174 0.0002 -0.0314 0.0046 0.0045 0.0039 0.0049 1 capex 0.3281 0.0102 -0.1156 -0.0066 -0.0113 0.0917 -0.1674 0.0266 0.0627 0.0607 0.008 1 𝑭𝑭𝑭𝑭𝑭𝑭𝟑𝟑𝟑𝟑 0.1169 0.0275 -0.041 -0.0026 -0.0001 0.0255 -0.0162 -0.0003 -0.1472 -0.2453 -0.0221 0.0268 1 𝑭𝑭𝑭𝑭𝑭𝑭𝟏𝟏 0.0707 0.0019 0.0041 -0.0057 -0.0062 0.0623 0.0042 0.022 -0.0117 -0.0686 0.0072 0.0827 0.003 1 GDP -0.158 0.012 0.0186 -0.0031 0.0135 0.0028 -0.035 -0.0291 0.0066 0.1426 -0.0083 -0.0559 -0.0446 -0.468 1

15 / 35

foreign exchange exposure and their cash holdings while there will be a statistically significant difference in the coefficients for foreign exchange exposure between financially constrained and constrained firms and the one for constrained firms is larger.

Part IV. Results

Figure II shows the trend of fluctuations of average cash holdings from the year of 2000 to 2015.It can be concluded that cash ratio follows a generally increasing trend with great fluctuations from 2005 to 2009. The fluctuations may be explained by Figure III, which represents the average cash and marketable securities and book value of total asset. It can be inferred from Figure III that the fluctuations from the year of 2005 to 2009 is mainly caused by the rise in the denominator part, total asset. Generally speaking, it can be concluded from Figure I that cash holding of nonfinancial firms keeps rising gradually until 2005 and then drops. It can also be observed from the figure that the rising trend of cash holdings becomes weaker after the financial crisis year of 2009.Such an increase in the year of 2009,which is only one year after the financial crisis in 2008,may be caused by a decrease in the origination of bank lending for real investment which has been reduced by 14% ( Ivashina&Scharfstein,2010).However, even firms have drawn down their credit lines in order to guarantee their liquidity needs, to ensure their future investment financing, both financially constrained and unconstrained firms have still reported an increase in cash reserve and a decrease in capital investment in their planned policies (Campello ,Graham & Harvey ,2010).In their research, they have concluded as well that due to the shortage in external finance, the increase in cash reserve financially constrained firms will be larger than that for unconstrained firms as they have already had difficulty in access to capital markets. Last but not least, after the year of 2010, the cash holdings of nonfinancial firms start to decrease slowly as the macroeconomic surroundings is slowly recovering from the financial crisis, supported by the increasing GDP shown in Figure I.

Figure II. Average Cash Ratios by Years

16 / 35 utility (SICs 6000 to 6999) from the year of 2000 to 2015.Firms with negative asset or sales and those firms with foreign income to net income less than 20% are removed as well. The cash ratio will be measured as cash and short-term investment to net asset (cash ratio)

Figure III. Average Cash and Marketable Securities, Net Asset and Total Asset by Year

This figure displays the average cash and marketable securities, net asset and total asset for firms exclude financial (SICs 4900 to 4999) and utility (SICs 6000 to 6999) from the year of 2000 to 2015.Firms with negative asset or sales and those firms with foreign income to net income less than 20% are removed as well. The two variables show the change of the denominator and nominator of the cash holding measure shown in Figure I.

17 / 35

firms under different criteria from 2000 to 2015.In Figure IV,Graph (1) and (2) are average cash ratios for firms when the criteria is dividend payout ratio and asset tangibility while Graph (3) presents the means of average cash ratios when firms are seperated based on the KZ index.As can be seen from Figure IV,generally

speaking,financially constrained firms have a much higher cash reserve than firms easily access to capital market.This result is almost the same as most previous studies that constrained firms hold more cash to face their future investment opportunities as it is costly for them to obtain financial support from external investors.However,when KZ index is regarded as the standard to separate constrained firms from unconstrained ones,during the period from the year of 2010 to 2012,financially unconstrained firms have a higher anverage cash ratio than those which are difficult to raise capital support in external capital market.

Figure IV. Average Cash Ratios for Constrained and Unconstrained Firms under Different Criteria by Years

This figure displays the average cash ratios for firms exclude financial (SICs 4900 to 4999) and utility (SICs 6000 to 6999) from the year of 2000 to 2015.Firms with negative asset or sales and those firms with foreign income to net income less than 20% are removed as well. Graph (1) and (2) show the average cash ratio when dividend or asset tangibility are regarded as the financial constraint criteria while Graph (3) are the average of cash ratio when firms are grouped by KZ index.

18 / 35

Table III presents regression estimations based on equation (2) for monthly and daily foreign exchange exposure. Column (1) in table III shows the regression result for cash ratio on monthly foreign exchange exposure while Column (2) displays the OLS regression on daily foreign exchange exposure.

In Column (1), which includes annual foreign exchange exposure derived from monthly observations, it can be concluded that when the firm’s return is more exposed to the change in trade-weighted US dollar currency exchange rate index when controlled by market return, the firm will not increase its cash holding in order to hedge itself against the risk brought by the exchange rate volatility. Therefore, the positive and statistically insignificant coefficient of the independent variable FXE does not support the first hypothesis that firms which are more deeply exposed to foreign exchange risk will reserve more cash.

As for other firm-specific estimators, several results originated from previous studies are confirmed. When firms are expected to obtain more future investment opportunities, they are more possible to have a high market-to-book ratio and from Column (1) in Table III, they will prefer to have a high cash and marketable securities to total asset ratio because it faces more growth opportunities, which is usually together with a high market-to-book ratio. For firms with more research and development expenses, which are also regarded to have more potential future growing opportunities, they may also increase their cash holdings relative to their ratio of research and development expenses to total sales. Besides, the coefficient of hedge performance in Column (1) in Table III indicates that derivatives held by nonfinancial firms in order to hedge themselves from foreign exchange exposure does not play an important role in the determination of their cash holdings. Different from the results obtained by Jong, Ligterink and Macrae (2006), on-balance derivatives fail to reduce the foreign exchange exposure.

Table III.The regression results for cash holdings

This table represents the regression results for cash to asset (equation (2)). The sample firms exclude financial (SICs 4900 to 4999) and utility (SICs 6000 to 6999) firms and the sample period is 2000 to 2015. Firms with negative asset or sales and those firms with foreign income to net income less than 20% are removed as well. Variable

19 / 35 definitions are provided in the Appendix.

(1) (2)

VARIABLES Cash ratio Cash ratio

Asset tangibility 0.446*** 0.445***

(0.0285) (0.0284)

Foreign income -1.82e-05 -1.64e-05

(2.02e-05) (2.08e-05) Acquisition to asset -0.0381*** -0.0387*** (0.0144) (0.0145)

dividend -4.43e-05 -4.94e-05

(0.000101) (0.000101)

RD sales 8.07e-05*** 8.11e-05***

(5.88e-06) (5.99e-06) leverage -0.0684*** -0.0628*** (0.0187) (0.0190) Nwc to asset 0.0788*** 0.0796*** (0.0251) (0.0250) q 0.00858*** 0.00867*** (0.00268) (0.00264) Cash flow to asset -0.0366** -0.0340* (0.0186) (0.0190) Firm size 0.0154* 0.0156* (0.00825) (0.00824) Hedge performance -0.000462 -0.000392 (0.000808) (0.000809) capex -0.525*** -0.506*** (0.0612) (0.0610) 𝐹𝐹𝐹𝐹𝐶𝐶30 0.000391 (0.000330) GDP 0.0746*** 0.0602*** (0.0147) (0.0147) 𝐹𝐹𝐹𝐹𝐶𝐶1 0.0999*** (0.0217) Constant -0.985*** -0.847*** (0.113) (0.113) Observations 8,339 8,341 R-squared 0.879 0.879

Firm FE YES YES

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Completely opposite to the study conducted by Almeida, Campello and Weisbach (2004) which says that when there comes a negative macroeconomic shock, the cash holdings of nonfinancial firms will rise, the regression results in Table III indicates that

20 / 35

when the GDP in the last year increases, managers of nonfinancial firms want to save more cash from the firms’ annual cash flows. The reason why managers take such action may be that they think that when the economic surroundings in the last year get better, there will be more potential investment opportunities in the next several years which require enough cash for financial support. There is also another explanation of such a situation that managers want to prevent the firms from cash shortage for possible future financial crisis so that during years with a better economic condition. During the years with good economic conditions, a greater cash inflow can be expected so that firms can reserve more cash in their cash flows to face up with future possible fund shortage.

In Column (2), the ratio of cash and marketable securities to total asset is regressed on annual foreign exchange exposure which is generated from daily returns and changes in trade-weighted US dollar currency exchange index. It can be concluded that similar results can be obtained except several coefficients. Among the several differences, there lies the key elements that the paper focuses on: the extent of foreign exchange exposure. According to the regression estimations based on equation (2) presented in Column (2) of Table III, the extent of foreign exchange exposure seems to have a significant and positive impact on the cash holdings of nonfinancial firms. In both Column (1) and (2), capital expenditure is proven to have a statistically significant negative influence on non-financial firms’ cash holdings, which indicates that for firms which spend more to create assets that can be viewed as collateral when getting access to external capital market, they hold less cash as they may have enough debt capacity.

Table IV and V display regression estimation for the two measures of foreign exchange exposure over different periods (daily and monthly), taking different measures to separate sample into financially constrained and unconstrained firms. Column (1) and (2) are under the criteria of dividend payout ratio while Column (3) and (4) show the results taking asset tangibility as the financial constraints criteria. Then the last two Columns (Column (5) & (6)) present the regression results based on equation (4) if firms are divided according to KZ index which is defined by equation (3).

21 / 35

In Table IV, most of the coefficient estimators are similar to the results in Table III. However, there are some differences between the demand function for cash holdings for unconstrained and constrained firms as well. It seems that constrained firms concentrate on the fraction of foreign income to their net income as well as their foreign exchange exposure generated from their return, market return and percentage change in trade-weighted US dollar currency exchange rate index except when the financial constraint criterion is determined as dividend payout ratio. Column (1)，(3) and (5) report that when constrained firms are more deeply exposed to fluctuations in US dollar exchange rate, they are more possible to hold more cash from their annual cash flows.

As for firm-specific determinants, compared with the regression results based on all sample firms in Table III, there is a difference in the coefficients of firm size found. With the entire sample, it can be inferred that larger firms are more willing to hold less cash while firms are separated as financial constrained and unconstrained ones, a completely opposite relationship is found no matter whether the firm is financial constrained or not, which means that larger firms want to hold more cash.

When a firm has a large number of cash flow, no matter whether it is financially constrained or not, it may save less cash as its managers may think that the firms is in a good condition so that more cash should be invested rather than reserved. The same as the conclusion originated from Table III, the performance of foreign exchange derivatives seems to have no significant influence on the cash holdings for financially constrained and unconstrained firms, which indicates that actions taken by nonfinancial firms’ managers so as to hedge the firms against foreign exchange risk do not work effectively. There only lies on exception for constrained firms under the criterion of asset tangibility that when such firms perform better in foreign currency derivatives hedging, they will reserve more. Even though the regression results are so similar to the ones from the whole sample, there is a difference in the coefficient of acquisition to asset. When all of the nonfinancial firms are divided into two subgroups, expenses related to acquisition activities have a significantly negative influence on the firms’ cash, meaning that no matter whether the firm has difficulty in access to external financing

22 / 35 Table IV. Regression Results for Financially Constrained and Unconstrained Firms

under Different Criteria

This table represents the regression results for cash to total asset on 𝐹𝐹𝐹𝐹𝐶𝐶₃₀(equation (4)). The sample firms exclude financial (SICs 4900 to 4999) and utility (SICs 6000 to 6999) firms and the sample period is 2000 to 2015. Firms with negative asset or sales and those firms with foreign income to net income less than 20% are removed as well. All the nonfinancial firms will be ranked based on their dividend payout ratio and asset tangibility respectively and those in the top(bottom) two quantiles of the total distribution will be assigned as financially constrained. Variable

definitions are provided in the Appendix.

Financial Constraint Criteria

Dividend payout ratio Asset tangibility KZ index

constrained unconstrained constrained unconstrained constrained unconstrained

VARIABLES (1) (2) (3) (4) (5) (6)

foreign income 0.000411 -3.17e-05 0.000413** -4.00e-05 0.00290*** -2.95e-05 (0.000803) (5.25e-05) (0.000175) (5.52e-05) (0.000781) (4.85e-05) acquisition to asset -0.219*** -0.190*** -0.0883*** -0.138 -0.193*** -0.175*** (0.0249) (0.0332) (0.0198) (0.102) (0.0288) (0.0350) RD sales 0.843*** 0.000163** 0.000198*** 0.00459*** 0.00248** 0.000168**

(0.0319) (7.17e-05) (5.22e-05) (0.00105) (0.00111) (6.57e-05) leverage -0.192*** -0.261*** -0.256*** -0.212*** -0.356*** -0.177*** (0.0108) (0.0121) (0.00923) (0.0158) (0.0122) (0.0124) nwc to asset 0.0945*** 0.0773*** 0.0529*** 0.0413*** 0.0892*** 0.0580*** (0.0120) (0.0131) (0.0116) (0.0156) (0.0126) (0.0141) q 0.0350*** 0.0413*** 0.0322*** 0.0528*** 0.0515*** 0.0384*** (0.00178) (0.00163) (0.00144) (0.00194) (0.00179) (0.00165) cash flow to asset -0.204*** -0.242*** -0.126*** -0.235*** -0.286*** -0.276*** (0.0373) (0.0166) (0.0216) (0.0193) (0.0239) (0.0178) firm size -0.00758*** -0.00726*** -0.00817*** -0.0118*** -0.0128*** -0.00251 (0.00101) (0.00156) (0.00104) (0.00154) (0.00113) (0.00168) hedge performance 0.000324 -0.00125 0.00891** -0.00200 -0.00874 -0.000749 (0.00919) (0.00190) (0.00420) (0.00639) (0.00644) (0.00178) capex -0.207*** -0.411*** -0.609*** -0.639*** -0.478*** -0.315*** (0.0421) (0.0483) (0.0611) (0.0490) (0.0479) (0.0483) 𝐹𝐹𝐹𝐹𝐶𝐶30 0.00202** 0.000560 0.00293*** 0.000475 0.00406*** 0.000909 (0.000824) (0.000671) (0.000697) (0.000771) (0.000877) (0.000666) GDP 0.109*** 0.0391*** 0.106*** 0.0752*** 0.0883*** 0.0562*** (0.00928) (0.0144) (0.0101) (0.0162) (0.0114) (0.0148) Constant -0.876*** -0.115 -0.790*** -0.434*** -0.573*** -0.345** (0.0882) (0.137) (0.0956) (0.154) (0.108) (0.140) Observations 3,860 4,479 4,946 3,393 4,900 3,439 R-squared 0.441 0.290 0.291 0.345 0.390 0.270

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

23 / 35

long as it is defined by dividend payout ratio. However, for unconstrained firms which are defined by their high asset tangibility, there is an insignificant relationship between cash holdings and acquisition to asset.

As for the changes in the macroeconomic environment, such as Gross Domestic Product(GDP), there still lies a positive relationship between change in GDP and nonfinancial firms’ cash holdings. In addition, it can be found when comparing Column (1) and (2), or Column (3) and (4), financially constrained firms react more sensitively than those unconstrained due to a larger coefficient of GDP discovered in the OLS regression based on equation (4).

Different from Table IV, which displays the grouped regression result of ratio of cash and marketable securities to total asset on monthly foreign exchange exposure, Table V shows the grouped regression result of ratio of cash and marketable securities

Table V. Regression Results for Financially Constrained and Unconstrained Firms under Different Criteria

This table represents the regression results for cash to total asset on 𝐹𝐹𝐹𝐹𝐶𝐶₁ (equation (4)). The sample firms exclude financial (SICs 4900 to 4999) and utility (SICs 6000 to 6999) firms and the sample period is 2000 to 2015. Firms with negative asset or sales and those firms with foreign income to net income less than 20% are removed as well. All the nonfinancial firms will be ranked based on their dividend payout ratio and asset tangibility respectively and those in the top(bottom) two quantiles of the total distribution will be assigned as financially constrained. Variable

definitions are provided in the Appendix.

Financial Constraint Criteria

Dividend payout ratio Asset tangibility KZ index

constrained unconstrained constrained unconstrained constrained unconstrained

VARIABLES (1) (2) (3) (4) (5) (6)

Foreign income 0.000587 -2.95e-05 0.000410** -3.83e-05 0.00310*** -2.68e-05

(0.000800) (5.24e-05) (0.000175) (5.52e-05) (0.000781) (4.85e-05) Acquisition to asset -0.218*** -0.190*** -0.0884*** -0.138 -0.194*** -0.176***

(0.0249) (0.0332) (0.0199) (0.102) (0.0289) (0.0349)

RD sales 0.849*** 0.000162** 0.000193*** 0.00458*** 0.00268** 0.000166**

(0.0318) (7.17e-05) (5.22e-05) (0.00105) (0.00111) (6.57e-05)

leverage -0.190*** -0.259*** -0.252*** -0.211*** -0.355*** -0.173*** (0.0108) (0.0120) (0.00927) (0.0157) (0.0122) (0.0124) Nwc to asset 0.0935*** 0.0771*** 0.0527*** 0.0403*** 0.0864*** 0.0575*** (0.0120) (0.0131) (0.0116) (0.0155) (0.0126) (0.0141) q 0.0349*** 0.0414*** 0.0324*** 0.0528*** 0.0515*** 0.0385*** (0.00178) (0.00163) (0.00144) (0.00194) (0.00180) (0.00165)

24 / 35

Cash flow to asset -0.203*** -0.243*** -0.135*** -0.236*** -0.289*** -0.278***

(0.0373) (0.0166) (0.0216) (0.0193) (0.0239) (0.0178) Firm size -0.00809*** -0.00755*** -0.00920*** -0.0120*** -0.0139*** -0.00294* (0.000990) (0.00153) (0.00101) (0.00150) (0.00110) (0.00166) Hedge performance -0.000173 -0.00120 0.00869** -0.00192 -0.00890 -0.000710 (0.00919) (0.00190) (0.00421) (0.00639) (0.00645) (0.00178) capex -0.197*** -0.405*** -0.592*** -0.637*** -0.465*** -0.309*** (0.0420) (0.0484) (0.0612) (0.0491) (0.0480) (0.0484) 𝐹𝐹𝐹𝐹𝐶𝐶1 0.0912*** -0.0783 0.0881** -0.0470 -0.0872* -0.0649 (0.0325) (0.0487) (0.0356) (0.0538) (0.0498) (0.0397) GDP 0.0961*** 0.0268* 0.0946*** 0.0672*** 0.0795*** 0.0420** (0.0104) (0.0163) (0.0112) (0.0186) (0.0127) (0.0167) Constant -0.745*** 0.00904 -0.660*** -0.353** -0.469*** -0.200 (0.0994) (0.155) (0.106) (0.178) (0.122) (0.159) Observations 3,861 4,480 4,947 3,394 4,901 3,440 R-squared 0.442 0.290 0.289 0.345 0.388 0.270

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

to total asset on daily foreign exchange exposure based on the same financial constraint criteria as in Table V. Firstly, the same as the second hypothesis that financially constrained firms are more sensitive to the fluctuations in the trade-weighted US dollar currency exchange rate index, there is a positive relationship between foreign exchange exposure and cash holdings of constrained firms while the relationship is insignificant for unconstrained firms.

There are also difference in the coefficients of foreign income and firm size for grouped sample compared with the whole sample. Except the case that the financial constraint criterion is dividend payout ratio, constrained firms with a higher ratio of foreign income to net income will result in a higher cash holding level. The same as the regression results in Table IV, smaller firms tend to have a lower cash ratio, totally opposite to the coefficients of firm size based on the whole sample.

Part V. Robustness Check

There are several robustness checks estimated in order to address potentially possible concerns about model specification and other issues. In Table VI, OLS

25 / 35

regressions of the cash holdings on monthly and daily foreign exchange exposures, are presented.

In Column (1) and (3), only manufacturing firms (SIC from 2000 to 3999) are included in the sample, which is much smaller than the original sample which included all nonfinancial firms while in Column (2) and (4), the sample period becomes smaller, from 2000 to 2007.Generally speaking, the sample pattern of the demand function for cash holdings can be argued. When the sample is not separated into financially constrained and unconstrained ones and only manufacturing firms are taken into consideration, expenses related to firms’ acquisition activities do not play a significant role in the determination of cash holdings.

Most importantly, a positive and significant relationship between foreign exchange exposure based on daily returns and firms’ cash holdings is confirmed again with an alternative sample period(Column (2) and (4)), which supports the first hypothesis made in this paper. Similarly, when an alternative sample which only contains manufacturing firms is applied (Column (1) and (3)), the robustness of the regression model is not challenged as well due to the statistically insignificant coefficient for foreign exchange exposure.

Compared with the robustness check results with an alternative sample, the robustness of the ones resulting from an alternative sample period. From Column (1) and (3), only the coefficients of foreign income, dividend payout ratio, cash flow to asset and hedge performance have experienced changes when only regressing with manufacturing firms. On the other hand, with only observations before the financial crisis in the year of 2008, only the sign and significance of asset tangibility, RD sales, firm size, hedge performance, capital expenditure, GDP as well as the key variable in the regression model, annual foreign exchange exposure derived from daily observations remain the same as the ones for the whole sample period, from 2000 to 2015.

Table VI. Robustness Check

This table shows results based on equation (2) for alternative sampling and period. The sample firms exclude financial (SICs 4900 to 4999) and utility (SICs 6000 to 6999) firms and the sample

26 / 35 period is 2000 to 2015. Firms with negative asset or sales and those firms with foreign income to net income less than 20% are removed as well. Column (1) and (2) are the result based on an alternative sample consists of only manufacturing firms while Column (3) and (4) show the robustness check with an alternative time period from 2000 to 2007.

(1) (2) (3) (4)

VARIABLES Cash ratio Cash ratio Cash ratio Cash ratio Asset tangibility 0.471*** 0.551*** 0.468*** 0.547*** (0.0334) (0.0483) (0.0333) (0.0481) Foreign income 0.000156** -0.000194* 0.000161*** -0.000205**

(6.64e-05) (0.000105) (6.21e-05) (9.58e-05) Acquisition to asset -0.0342** -0.0396 -0.0344** -0.0406

(0.0152) (0.0290) (0.0153) (0.0290) dividend 0.000157*** -0.000157* 0.000157*** -0.000160**

(5.57e-05) (8.47e-05) (5.25e-05) (7.56e-05) RD sales 8.23e-05*** -0.00185*** 8.32e-05*** -0.00186**

(6.96e-06) (0.000703) (7.01e-06) (0.000744) leverage -0.0563*** -0.0452 -0.0515** -0.0430 (0.0195) (0.0320) (0.0200) (0.0319) Nwc to asset 0.0637** 0.0368 0.0643** 0.0396 (0.0305) (0.0397) (0.0305) (0.0397) q 0.00588* 0.00353 0.00595* 0.00448 (0.00347) (0.00370) (0.00341) (0.00371) Cash flow to asset -0.0393 -0.0375 -0.0345 -0.0334

(0.0357) (0.0409) (0.0358) (0.0406) Firm size 0.0320*** 0.0390*** 0.0323*** 0.0407*** (0.0113) (0.0120) (0.0113) (0.0120) Hedge performance 0.00364** 0.00122 0.00380** 0.00122 (0.00164) (0.00223) (0.00155) (0.00210) capex -0.663*** -0.510*** -0.632*** -0.485*** (0.0894) (0.0932) (0.0886) (0.0930) 𝐹𝐹𝐹𝐹𝐶𝐶30 0.000428 0.000492 (0.000356) (0.000499) 𝐹𝐹𝐹𝐹𝐶𝐶1 0.0935*** 0.0821*** (0.0248) (0.0267) GDP 0.0696*** 0.0802*** 0.0561*** 0.0506* (0.0168) (0.0265) (0.0170) (0.0279) Constant -1.077*** -1.282*** -0.949*** -1.014*** (0.123) (0.218) (0.126) (0.230) Observations 5,690 3,833 5,690 3,833 R-squared 0.876 0.921 0.877 0.922

Firm FE YES YES YES YES

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

27 / 35

that financially constrained firms are more sensitive to the fluctuations in foreign exchange exposure, two different robustness checks will be applied in the regression. As this paper mainly concerns about the effect of foreign exchange exposure on firms’ cash holding level, in order to simplify the expression of the robustness check, only the regression results for the main variable FXE are shown in Table VII.

As can be observed from Table VII, with such an alternative sample, it can be claimed that regression model including 𝐹𝐹𝐹𝐹𝐶𝐶₃₀ is more robust than the model with 𝐹𝐹𝐹𝐹𝐶𝐶1 as when the annual foreign exchange exposure is measured by monthly returns

and currency exchange rate index, cash holdings of financially constrained firms are more sensitive to their extent of foreign exchange exposure when they are

distinguished by asset tangibility or KZ index, while if dividend payout ratio is the criterion to classify financially constrained firms, the robustness of the regression model is also challenged, just as the cases when foreign exchange exposure is measured by daily observations no matter which financial constraint criteria is applied.

Table VII. Robustness Check for Grouped Sample

This table shows results based on equation (4) for alternative sampling and period. The sample firms only include manufacturing firms (SIC from 2000 to 3999) and the sample period is 2000 to 2015. Firms with negative asset or sales and those firms with foreign income to net income less than 20% are removed as well. All the nonfinancial firms will be ranked based on their dividend payout ratio, asset tangibility respectively and KZ index (defined in equation (3)). To save space, only the coefficients and standard errors for foreign exchange exposure under each condition are presented.

Financial Constraint Criteria

Dividend payout ratio Asset tangibility KZ index Manufacturing firms 𝐹𝐹𝐹𝐹𝐶𝐶₁ 𝐹𝐹𝐹𝐹𝐶𝐶₃₀ 𝐹𝐹𝐹𝐹𝐶𝐶₁ 𝐹𝐹𝐹𝐹𝐶𝐶₃₀ 𝐹𝐹𝐹𝐹𝐶𝐶₁ 𝐹𝐹𝐹𝐹𝐶𝐶₃₀ constrained 0.0542 0.00037 0.0505 0.00340*** 0.0313 0.00346*** (0.0359) (0.000984) (0.038) (0.00077) (0.0443) (0.00103) unconstrained 0.0548 0.000707 0.031 0.000519 0.0657 0.000346 (0.0609) (0.000748) (0.0678) (0.00087) (0.0602) (0.00073) Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Other than the robustness with an alternative sample which only consists of manufacturing firms, another robustness check can be run with an alternative financial

28 / 35

constraint criterion: Whited-Wu index，which is determined by Whited and Wu in their study conducted in 2006.As in their study, the index is defined as the following equation:

𝑊𝑊𝑊𝑊 = − 0.091𝐶𝐶𝑎𝑎𝑎𝑎ℎ𝑓𝑓𝐹𝐹𝐹𝐹𝐹𝐹 − 0.062𝐺𝐺𝐹𝐹𝐿𝐿 + 0.021 ∗ 𝑇𝑇𝐿𝐿𝑇𝑇𝐺𝐺 − 0.044 ∗ 𝑓𝑓𝐹𝐹𝐹𝐹𝐹𝐹𝑎𝑎𝐹𝐹𝑓𝑓𝑎𝑎 + 0.102 ∗ 𝐼𝐼𝐼𝐼𝐺𝐺 − 0.035 ∗ 𝐼𝐼𝐺𝐺 (5) Where Div is a dummy variable which is equal to 1 if the firm pays out dividends in this year. ISG is the sales growth based on the firm’s three-digit SIC industry code while SG is determined by the firm’s sale growth.

As can be observed from Table VII which represents the robustness check results when WW index is regarded as the financial constraint criterion, the second hypothesis that financially constrained firms react more greatly than unconstrained ones to fluctuations in exchange rate is still supported with another financial constraint criterion. No matter how the degree of foreign exchange exposure is

measured, based on daily or monthly observations, financial constrained firms tend to increase more cash holdings when their return is more related to changes in exchange rate of US dollar than those which have no difficulty in getting financed in external capital market. The robustness of the regression model is only challenged by the coefficient of research and development expenses. When firms are divided according to the corresponding WW index, the relationship between RD to sales and cash ratio becomes insignificant but still keeps positive.

Table VII. Robustness Check with Alternative Financial Constraint Criteria This table shows results based on equation (4) for alternative financial constraint criteria. The sample firms exclude financial (SICs 4900 to 4999) and utility (SICs 6000 to 6999) firms and the sample period is 2000 to 2015. Firms with negative asset or sales and those firms with foreign income to net income less than 20% are removed as well. All the nonfinancial firms will be ranked based on their WW index (defined in equation (5)).

Financial Constraint Criteria WW index

constrained unconstrained constrained unconstrained VARIABLES (1) (2) (3) (4) Foreign income 1.57e-05 -2.50e-05 -5.98e-06 2.69e-05

(4.05e-05) (0.000164) (4.07e-05) (0.000164) Acquisition to asset -0.180*** -0.211*** -0.179*** -0.211***

29 / 35 (0.0244) (0.0414) (0.0245) (0.0413) RD sales 0.00264 0.000151** 0.00244 0.000150** (0.00173) (7.47e-05) (0.00174) (7.47e-05) leverage -0.223*** -0.283*** -0.220*** -0.282*** (0.0102) (0.0145) (0.0103) (0.0145) Nwc to asset 0.0975*** 0.0728*** 0.0971*** 0.0721*** (0.0121) (0.0149) (0.0122) (0.0149) q 0.0490*** 0.0393*** 0.0489*** 0.0393*** (0.00152) (0.00200) (0.00153) (0.00200) Cash flow to asset -0.212*** -0.265*** -0.226*** -0.265*** (0.0290) (0.0184) (0.0290) (0.0184) Firm size -0.00379*** -0.00495** -0.00513*** -0.00521** (0.00117) (0.00212) (0.00116) (0.00209) Hedge performance -0.00181 0.000213 -0.00207 0.000243 (0.00527) (0.00408) (0.00529) (0.00408) Capex -0.304*** -0.527*** -0.277*** -0.524*** (0.0400) (0.0578) (0.0400) (0.0580) 𝐹𝐹𝐹𝐹𝐶𝐶30 0.00555*** 0.000453 (0.000780) (0.000749) 𝐹𝐹𝐹𝐹𝐶𝐶1 0.122*** -0.0323 (0.0366) (0.0538) GDP 0.107*** 0.0753*** 0.0917*** 0.0697*** (0.0102) (0.0166) (0.0113) (0.0190) Constant -0.874*** -0.463*** -0.700*** -0.406** (0.0975) (0.158) (0.109) (0.182) Observations 4,865 3,474 4,866 3,475 R-squared 0.326 0.279 0.320 0.279 Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Part VI. Conclusion

This paper mainly focuses on the influence of foreign exchange exposure on the

determination of cash holdings of nonfinancial firms. Previous studies (Jong, Ligterink and Macrae ,2006) have concentrated on its effect on firms’ cash flows rather than cash holding and have come up with the conclusion that foreign exchange exposure seems to have no significant impact on firms’ annual total cash flows due to actions taken by managers in order to hedge against the risk using on-balance derivatives rather than off-balance derivatives. While for the determination of firms’ cash holdings, based on

30 / 35

former research (Opleretal,1999; Almeida, Campello and Weisbach,2004; Bates, Kahle and Stulz,2009) which has been confirmed that financially constrained firms are more likely to hold more cash than those unconstrained regressions on nonfinancial firms’ cash holdings are applied to study the relationship between cash holding and foreign exchange exposure.

In conclusion, the effect of foreign exchange exposure on nonfinancial firms’ cash holdings differs with the determination of foreign exchange exposure. When the level of foreign exposure is scaled by monthly observations, fluctuation in exchange rate seems to play an insignificant but positive influence on the cash holdings of non-financial firms. On the other hand, if the degree of foreign exchange exposure is defined by daily corresponding data, the hypothesis that non-financial firms prefer to increase their cash holdings with the rise of their extent of foreign exposure. This result implies that for firms that are more exposed to changes in exchange rate, they will reserve more cash from their annual cash flows and the positive relationship is much greater for those nonfinancial firms who have difficulty in access to external capital market compared to those who can easily raise external funds.

Consequently, this paper does not support existed theory that derivatives

eliminate the impact of foreign exchange exposure on firms’ operations but is in line with the theory that firms hold more cash to face potential future investment opportunities. With more exposed to fluctuation of trade-weighted US dollar currency exchange rate index, more cash holdings are more desirable for firms so as to protect themselves from foreign exchange risk. The results of this paper are robust to firms in a certain time period but the robustness with an alternative sample is challenged. Additionally, when firms are divided to financially constrained and unconstrained ones, the robustness of the results derived from this paper is doubted. With the help of the conclusions of this paper, firms which are more exposed to foreign transactions and international competition can improve their risk management by adjusting their cash holding level.

31 / 35

not taken into consideration and whether the financial crisis from the year of 2008 affects the demand function of cash holdings of nonfinancial firms, which is mentioned by Pinkowitz, Stulz & Williamson (2013), is not included in this research as well. Therefore, further research can take efforts on industry-level effects and the impact of financial crisis on the foreign exchange exposure, thus changing firms’ cash holding levels.

32 / 35 Bibliography

Acharya, V., H. Almeida, and M. Campello, 2006, Is cash negative debt? A hedging perspective on corporate financial policies, London Business School working paper, London, England.

Adler, M., and Dumas, B., (1984) Exposure to Currency Risk: Definition and Measurement. Financial Management, Vol. 13, No. 2 (Summer, 1984), pp. 41-50 Almeida, H., Campello, M. and Weisbach, M. S. (2004), The Cash Flow Sensitivity of Cash. The Journal of Finance, 59: 1777–1804.

Almeida, H., Campello, M., (2007) Financial Constraints, Asset Tangibility, and Corporate Investment, The Review of Financial Studies, Volume 20, Issue 5, 1 September 2007, Pages 1429–1460

Allayannis, G., Ofek, E., 2001. Exchange rate exposure, hedging, and the use of foreign currency derivatives. Journal of International Money and Finance 20, 273–296.

Bates, T. W., Kahle, K. M. and Stulz, R. M. (2009), Why Do U.S. Firms Hold So Much More Cash than They Used To? The Journal of Finance, 64: 1985–2021.

Bartram, S. M. (2008) What lies beneath: Foreign exchange rate exposure, hedging and cash flows, Journal of Banking & Finance, Volume 32, Issue 8,2008, Pages 1508-1521

Brown, Gregory W (2001) Managing foreign exchange risk with derivatives, Journal of Financial Economics, Volume 60, Issues 2–3,2001, Pages 401-448,

Campello, M., Graham, J.R., Harvey, C.R. (2010). The real effects of financial constraints: evidence from a financial crisis. J. Finance. Econ. 97:470–87

Dittmar A., Mahrt-Smith, J., (2007) Corporate governance and the value of cash holdings, Journal of Financial Economics, Volume 83, Issue 3,2007, Pages 599-634 Dittmar A., Mahrt-Smith, J. and Servaes, H., (2003) The Journal of Financial and Quantitative Analysis, Vol. 38, No. 1 (Mar., 2003), pp. 111-133

Donnelly, R., Sheehy, E., 1996.The share price reaction of UK exporters to exchange rate movements: an empirical study. The Journal of International Business Studies 27 (First Quarter) 157–165.

33 / 35

Elliott, W.B., Huffman, S.P., Makar, S.D., 2003. Foreign-denominated debt and foreign currency derivatives: complements or substitutes in hedging foreign currency risk? Journal of Multinational Financial Management 13, 123–139.

Fazzari, S.R., Glenn, H. and Bruce, P. (1988) Financing constraints and corporate investments, Brooking Papers on Economic Activity 1,141-195

Fazzari, S., Hubbard, R., & Petersen, B. (1988). Investment, Financing Decisions, and Tax Policy. The American Economic Review, 78(2), 200-205. Retrieved from http://www.jstor.org.proxy.uba.uva.nl:2048/stable/1818123

Froot, Kenneth A Scharfstein, David S; Stein, Jeremy C (1993) Risk management: coordinating corporate investment and financing policies. The journal of finance: the journal of the American Finance Association, 1993, Vol.48(5), pp.1629-1658

Fritz Foley C., Jay C. Hartzell, Sheridan Titman, Garry Twite, (2007) Why do firms hold so much cash? A tax-based explanation, Journal of Financial Economics, Volume 86, Issue 3,2007, Pages 579-607

Griffin, John M., Stulz, René M. (2001) International Competition and Exchange Rate Shocks: A Cross-Country Industry Analysis of Stock Returns, The Review of Financial Studies, Volume 14, Issue 1, 1 January 2001, Pages 215–241

Han, Seungjin, and Jiaping Qiu, (2007), Corporate precautionary cash holdings, Journal of Corporate Finance 13, 43–57.

Huffman, S.P., Makar, S. D., (2004) The effectiveness of currency-hedging techniques over multiple return horizons for foreign-denominated debt issuers, Journal of Multinational Financial Management, Volume 14, Issue 2, Pages 105-115

Ivashina V., Scharfstein, D. (2010). Bank lending during the financial crisis of 2008. J. Financ.Econ.97:319–38

Jensen, M., 1986, Agency costs of free cash flow, corporate finance and takeovers, American Economic Review 76, 323-329.

Jong, Abe; Ligterink, Jeroen; Macrae, Victor (2006) A Firm-Specific Analysis of the Exchange-Rate Exposure of Dutch Firms. Journal of International Financial Management and Accounting, 04/2006, Vol.17(1), pp.1-28

34 / 35

Kaplan, S. and Zingales, L., (1997) Do financing constraints explain why investment is correlated with cash flow? Quarterly Journal of Economics 112,169-215

Lamont, O., Polk, C., & Saá-Requejo, J. (2001). Financial Constraints and Stock Returns. The Review of Financial Studies, 14(2), 529-554.

Lane, Philip R.; Milesi-Ferretti, Gian Maria (2005) Financial Globalization and Exchange Rates. Working Paper No. 05/3

Riddick, Leigh A., and Toni M. Whited, (2009), The corporate propensity to save, Journal of Finance 64, 1729–1766.

Opler T, Pinkowitz L, Stulz RM, Williamson R. 1999. The determinants and implications of corporate cash holdings. J. Finance. Econ. 52:3–46

Sudacevschi, M. (2017) Foreign currency risk hedging, Challenges of the Knowledge Society, 7(-):742-746

Whited, Toni M., Wu, Guojun; (2006) Financial Constraints Risk, The Review of Financial Studies, Volume 19, Issue 2, 1 July 2006, Pages 531–559