The effect of product market competition on cash holdings

(1)

The effect of product market competition on cash holdings

University of Amsterdam

Master Thesis

MSc Finance Quantitative Finance

Student Name: He Tian

Month and Year: July, 2018

Supervisor: R. Almeida Da Matta

(2)

Statement of Originality

This document is written by Student He Tian, 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)

Abstract:

The aim of this paper is to investigate the relationship between product market competition and firm cash holdings of all publicly traded manufacturing firms in the US over 2002-2016. This paper finds that the effect of industry competition intensity on firm cash holdings is not monotonous. In a very competitive environment, firms’ cash holdings increase if the

competition becomes more intense, whereas less competition results in a higher cash holding if firms are operating in the more general and very monopoly industries. The above effects remain the same direction after controlling firm position within the corresponding industry. Moreover, Firms which have median productivity level within the industry tend to hold more cash.

(4)

1.Introduction

Cash plays a crucial role in the company governance since the capital market is not perfect in the real world, raising funds can be costly sometimes. Not only can cash be used as a tool to provide liquidity payments for business operations, but it can also provide financing sources for investment activities to capture market opportunities. Moreover, firms would like to stock cash for the precautionary reason and increase their ability to address

unexpected shock as well. Holding cash is a strategy of the firm and the level of cash stock is an important factor in company governance. Much attention has been devoted to

investigating the determinants of firm cash holding level based on three famous models, trade-off, pecking order and free cash flow model respectively. These three modules focus on the environment inside the company. However, the corporate decisions should not be made by only considering firm’s own situation, the external governance mechanism, which is product market competition, matters as well.

Product market competition is the bridge between macroeconomics and micro-firm. Main interests of competition come along with capital structure. The literature that

corresponding to cash holding is rare. The effect of competition on firm cash holding level is debatable among previously limited literature. There are two main corresponding

arguments. Some people argue that the profitability of the industry will decrease because of the higher competition intensity, thus firms which are in the competitive environment are more likely to suffer the loss and firm would like to stock more cash for the precautionary reason and competitive pressure. Morellec, Nikolov and Zucchi (2009) found that higher competition level is associated with higher cash holding level, they pointed out the optimal cash stock level increase when the competition is intense as well. However, some people consider that firms that in the monopoly industry face a higher chance of predatory behavior and firms are intend to stock extra cash in order to avoid predatory risk. The empirical results obtained by Haushalter, Klasa and Maxwell (2007) confirmed this hypothesis.

The above two arguments both make sense. However, either of them is looking at the whole picture of the competition. Morellec, Nikolov and Zucchi (2009) only considered the

(6)

side of the competition effect, without paying attention to the predatory risk that the firm could face when there are only a short number of participants. While Haushalter, Klasa and Maxwell (2007) did focus on the predatory behavior and came to the conclusion that cash holding level is higher in the monopoly industry, but they neglected the positive effect of competition on firms cash stock. This paper divides industries into several groups with respect to their competition intensity and attempt to make an analysis on the factors that firm will consider and the risks that firm will encounter corresponding to different range of industry competition levels, instead of jumping to the conclusion such as higher competition is associated with higher cash holdings and vice versa. Additionally, this paper will combine firm position(intra-competition) with industry competition intensity(inter-competition) as the proxy and test how these factors affect firm cash holdings simultaneously, and how would the effect of industry competition level on cash holding change after controlling the firm position effect.

This paper will analyse the cash holding level of the manufacturing firms in the US over 2002-2016. This sample has been selected for the following considerations. The validity of empirical evidence differs across different countries. The legal environment of the different country could also influence firm cash holding level. Siems(2008) indicated that developed country perform shareholder protection better than the developing countries. And the country has better shareholder protection would have less agency problem, which is a factor that affects firm cash holding level other than the main interest of this paper. Therefore, I narrowed down my research country to developed country. The reason that I chose the US is the cash ratio has been increased dramatically for US firms but lots of research researched its reason only based on firm’s internal environment. Therefore, it would be interesting to use US firms as the sample to test the relationship between competition and cash holdings.

This paper contributes in three ways. Firstly, it distinguishes the different effects of the different range of competition level on firm cash holding level instead of considering all industries as a whole. Secondly, a new measurement to capture the firm position within the industry is developed. Finally, even the main interest of this paper is the product market competition, other factors that are proven to influence cash holdings from previous

(7)

change with considering the product market competition compared with literature that only focus on the inside environment of the company are discussed in this paper as well.

This paper is designed in the following manner. Section 2 reviews the related literature, with showing the similarities and differences between previous literature and this paper, the hypothesis is presented in this section as well. Section 3 shows the data selection details and the main variables’ construction method, followed by a session with an econometric

framework which is designed to test the hypothesis and the expectations of the effects of main variables on cash ratio. Data descriptive statistics is present in section 5. Section 6 reports the empirical findings which obtained from the regression models, followed by the robustness checks session. Finally, the last section concludes the whole paper and presents the limitation.

2.Literature review and hypothesis

2.1 Overview of the previous literature

There is a large number of articles have analyzed the determinants of cash holdings. As mentioned in the previous session, most of them are corresponding to the firm

characteristics. It would be meaningful to review the corresponding literature even the main interest of this paper is the effect of product market competition. The research from Opler, Pinkowitz, Stulz and Williamson (1998) shows the determinants of firm’s cash holding. Their results are consistent with the trade-off theory, which is a point of view of maximizing shareholders’ wealth with considering the benefits and cost of holding cash. However, firms tend to stock more cash when they perform well instead of only considering maximizing shareholders’ welfare. Additionally, firms that have great access to the capital markets are reluctant to hold excess cash. The variables that they considered in the research are widely recognized as the influence of firm cash holding level. The variables other than competition intensity and firm position that have been used in this paper are followed theirs.

The average cash to assets ratio has been dramatically increased in US firms. Bates, Kahle and Stulz (2009) find that the precautionary motive is crucial to explain the incensement in cash stock level, while agency cost contributes less to this increase. Their findings are consistent with Opler, Pinkowitz, Stulz and Williamson (1998) as well. Lots of research

(8)

concludes that agency costs theory has less power to explain firm cash holding level. However, their paper did not analyze the effect of competition on firm cash holding level. With respect to competition, the main debates come from two different points of views of competition. Some people consider the competition level as the determinant of predatory risk, which is a crucial factor that firm will manage with holding extra cash. However, some people argue that firm tend to increase their cash holding level because of the competitive pressure and higher competition would decrease information asymmetry, resulting in a higher marginal benefit of holding cash. The following literature show the details of the above arguments.

A very comprehensive view of competition, cash holding, and investment behavior is

provided by Haushalter, Klasa, and Maxwell (2007). Their work mainly focused on the effect of predatory risk on firm cash holding level. The core of their work is the interdependence among firms, the more interdependence, the highly predatory risk that the firm will encounter, resulting in a higher cash holding level. Industry competition intensity and firm position within the industry are used to represent the predatory risk. They found that firms do manage predation risk, which is a crucial determinant of firm governance, predation risk is positively correlated with firm cash holding level. Their empirical results showed that firms which are in the relatively concentrated industry tend to hold more cash and firms that have median productivity compared with other rivals within the industry intend to have a higher cash ratio as well. However, they treated the competition among industries and within industry separately rather than considering the joint effect. The way to define the industry competition intensity in their paper was collecting Herfindahl-Hirschman

Index(HHI) from the Census of manufactures. The results are reported every five years. Therefore, the amount of HHI is the same every five years. This paper will use HHI to represent competition level using the same method.

The most recent research that focuses on predatory risk as well was provided by Yang and Zheng (2017). Instead of considering firm cash holding level, the marginal benefit of holding cash was the dependent variable. They argued that predatory risk pressure will push

managers to work hard and use cash more efficiently. The empirical result from Chinese listed firms’ situation confirms their hypothesis, which is the product market competition

(9)

could mitigate agency problems, thus increasing the marginal value of cash stock. However, their data come from Chinese industry, where the government decisions have huge impacts on firm decisions, thus whether their results are adapted in US firms are uncertain.

Instead of considering predatory risk, Morellec, Nikolov, and Zucchi (2009) combined industry competition intensity with trade-off theory. They figured firm cash holding level with respect to different HHI level over time using public traded firms in the US between 1980-2007, they found that firms that in the competitive industry have a higher cash holding level. Their empirical regression results confirm this positive correlation between

competition intensity and firm cash holding level as well. And this positive relationship is even more severe when the firm is in the financial constraint situation. Their paper used price to cost margin(EPCM) and HHI as two of the measurement to stand for competition without paying attention of the fact that HHI measures the competition among industries, whereas EPCM represents more about the competition within the industry. Based on the method and evidence of their paper, it would be necessary to distinguish the different kinds of competition. Therefore, this paper uses Herfindahl-Hirschman Index(HHI) to measure the industry competition intensity (inter competition) and firm position to represent the

competition within the industry (intra competition).

Another literature which is worth to mention is from Lyandres and Palazzo (2016), they combined the competition with investment behavior through innovative firms’ cash

holdings. They considered cash holding could stand for the likelihood of firms' investment in the future, which means higher cash holding level shows the higher possibility of firms' future investment when firms are in financial constraints situation. They argued that the level of cash holding negatively depends on rivals’ cash holding choices since the benefit of holding cash would decrease if their rivals intend to invest more. This negative relationship is stronger if the firm is facing more competition. Their consideration of competition among innovative firms is the degree of benefit that the investment could bring to firms, the marginal benefit of investing would be small when the competition is intense, thus firm will choose less cash holding. Moreover, he did not use the traditional way to define industry concentration level, instead, the number of patents and the number of citations to each patent were used to capture the innovation proximity of firm and its competitors. They used

(10)

this method mainly because they narrowed down his sample to innovating firms, so their method might be more efficient than the traditional HHI index.

As mentioned before, One of the main independent variables in this paper is firm position, the method of this variable construction will be followed the idea of Mackay(2005), who used the absolute value of the difference between firm’s technology ratio and the industry-year median technology ratio divided the range of technology ratio in each industry-industry-year. However, this paper will use total factor productivity instead as the measure to stand for firm’s position in the corresponding industry since firm’s technology level is not the only factor that determines firm’s position. More details about this firm position variable are present in the following session.

2.2 Hypothesis development

Based on the arguments of the previous literature, the effect of competition on cash holding is debatable. The positive effect of competition on cash holding comes from the intuition that higher competition could decrease information asymmetry and increase market transparency, which will increase the value of holding extra cash, and firm tend to hold more cash for the precautionary reason because of the competitive pressure. The negative effect is because the predatory risk happens more in the relatively monopoly industries and firms manage predatory risk with holding extra cash. It would not be wise to consider the single effect of competition on cash holdings. Based on the above points of view, the first hypothesis of this paper is:

H1: The correlation between industry competition intensity and cash holdings is not linear. When the competition is intense, firms’ motive of taking actions to obtain or keep

competitive power is more obvious. However, the influence of such actions is quite limited because of the large amounts of other rivals within the industry. Considering an extreme case, in a perfect competitive industry, there wouldn’t be any benefits to conduct predatory behavior since the interdependence among firms in the competitive industry is very low, there is no difference between a firm facing n-1 competitors and n competitors when n is quite large, the predatory behavior is very unlikely to happen. Therefore, the position effect

(11)

of competition on cash holding is dominant. This brings the second hypothesis of this paper:

H2: Higher industry competition intensity increase firm cash holding level among firms that are operating in the very competitive environment.

Compared to the industry where the competition is seriously intense, the interdependence is much higher in the monopoly environment. The possibility of the occurrence of predatory behavior becomes bigger, it is worthier for firms to conduct predatory behavior as well. Additionally, even the competition is less and rivals are rare in a monopoly environment, firms would feel more threaten by the predatory risk, resulting in higher benefits of holding extra cash to prevent predatory behavior. Therefore, the effect of predatory risk should be dominant compared to the opposite effect of the competitive pressure, which brings the third hypothesis of this paper:

H3: Higher monopoly results in a larger amount of cash holing among firms that are operating in the very concentrated industries.

As mentioned before, the more interdependence among firms, the higher possibility of the occurrence of predatory behavior, resulting a higher cash stock level. The interdependence among firms does not only depend on the competition among the various industries but also related to the competition within the industry. Firms’ total factor productivity is used as a proxy to represent whether the firm is in the core or in the fringe within the industry. If a firm is in the core of the industry, it means that this firm has more interdependence with its rivals since its productivity is the median of the industry, resulting in facing similar

investment opportunities with other firms in the same industry, thus the predatory risk is higher. So, the productivity of firms that are in the middle of the industry will tend to hold more cash to avoid underinvestment which leads to losing market share. Moreover, one thing that needs to emphasise is that the effect of firm position should be more severe if the firm is in the concentrated industry where the predatory risk is more likely to take place. Therefore, the other two hypotheses of this paper are:

(12)

H5: The above relationship becomes more severe if the firm is in the very concentrated industry.

3. Data

3.1 Data source and selection details

The sample in the paper is the manufacturing firms (NAICS between 311111 to 399999) in the US firm over the 2002-2016 period. All corresponding financial data of firms are obtained from Compustat-Capital IQ from Wharton database. Only the manufacturing sector is chosen is because one of the crucial variables in this paper is based on Cobb-Douglas production function. With respect to the industry code, both NAICS and SIC code could identify firm’s corresponding industry. However, the measurement of industry

competition intensity (HHI) are calculated from six-digit NAICS code and NAICS was updated to replace SIC code as well, thus NAICS code is used in this paper to identify firm’s industry. In this paper, observations with missing or non-positive total assets and sales are deleted. All independent variables except HHI are winsorized at 1 percent level in order to eliminate the effect of the outliners.

3.2 Industry competition intensity measurement

Herfindahl-Hirschman Index (HHI) is the most common proxy to measure industry

competition level. Higher HHI stands for lower competition. Its calculated as the sum of the squaring market share of each firm within the same industry. The six-digit NAICS industry concentration ratio is acquired from Census of Manufacturs for 2002, 2007, 2012 since US industry concentration ratio is reported every five years. The HHI data is collected from database instead of calculating based on its definition is due to the fact that only publicly traded firms’ data could be obtained from WRDS database, whereas HHI data from Census of Manufactures considers both publicly traded and private firms’ market share. Therefore, their result is more appropriate to stand for industry concentration. Additionally, the HHI amount is divided by 10000 for the better interpretation in the regressions since HHI can range from close to zero to 10000.

(13)

3.3 Firm position measurement

The determinants of the position of the firm is hard to decide. This paper mainly follows the idea from Mackay and Phillips (2005). Three factors have been considered in order to better measure the firm’s position within the industry.

Firstly, it should reflect firm’s productivity. The most common measurement of firm

productivity or production technology is using the capital to labor ratio, which only captures the technology level of the corresponding firm. This paper will use total factor productivity (TFP) instead. TFP refers to the portion output not explained by traditional inputs of labor and capital, thus this measurement not only captures the production technology of the firm but also contains some other factors that might have an influence on firm's position, policy factor for example. Compared to capital to labor ratio, total factor productivity represents more factors that could affect firm position within the industry. There is no fixed method to obtain total factor productivity, this paper calculates TFP using Cobb-Douglas production function.

Cobb-Douglas production function: Y= AKα Lβ, 0<α<1 and 0<β<1

The residual of the following equation stands for total factor productivity: Log(Yf i t)=α1 log(Kf i t) + α2 log(Lf i t)+ µf + µt +ɛf i t

Where Yis real sales, K stands for property plant and equipment. L is employment, µf and

µt represent firm and time fixed effect. f, i, t, stand for firm, industry and year respectively.

The residuals are firms’ total factor productivity.

Secondly, it should measure the distance between firm’s productivity and the median productivity level in the corresponding industry, whether it’s in the core or fringe within the industry. And the last consideration is the measurement should comparable among various industries.

(14)

Where mediani –f t is the median total factor productivity of the rest of firms (all firms

excludes the firm itself) for a given industry and year, thus the numerator stands for how much difference between the firm’s productivity and the rest of firms’ median productivity. Finally, divided by the range of TFP ratio deviations in each industry-year to make this measurement comparable. This firm position variable proxy between 0 and 1. A small index indicates the closer to industry median level. One thing that worth to mention is that this variable stands for the similarity between a firm and the rest of its industry, instead of showing whether the firm has a high or low total factor productivity since the numerator is the absolute value of the difference instead of the actual value. The smaller this proxy, the firm is more similar to the rest of its industry.

4. Methodology

4.1 Econometric framework

The econometric framework is as follow:

Cashf t = β1 HHIf t + β2 X f t + ɛ f t (1)

Cashf t = β1 HHIf t + β2 FP f t + β3 X f t + ɛ f t (2)

Where cash stands for cash ratio, which is (cash + short-term investment)/ assets. HHI and FP stand for Herfindahl-Hirschman Index (industry concentration) and firm position

respectively. X f t stands for control variables contain firm size, leverage, market to book

ratio, , R&D expenses/assets, net working capital/assets, acquisition/assets, capital expenditure/assets, divided dummy, and cash flow volatility. The main interests are the coefficient before HHI and FP. The first regression tend to test the first three hypothesis, whereas the second regression is used to test the last two hypothesis.

4.2 Expectation of the effect of competition

Based on the hypothesis in section 2, the coefficient before HHI is expected to be different among various groups. More specifically, the expectation of the correlation between HHI and cash ratio is negative in the very competitive group while the other way around in the very monopoly industries. For the effect of firm position, the expectation of the coefficient is negative, which is expected to be consist in all groups.

(15)

4.3 Prediction of the main control variables based on the theory

As mentioned before, the majority of cash holdings paper focus on firm characteristics using trade-off, pecking order and free cash flow model. This paper briefly summarizes the

conflicts of these models for the variables that have ambitious effect on cash ratios. Firm size: trade-off theory predicts a negative correlation between firm size and cash ratio because large firm could benefit from economies of scales, and raising funds is easier for large firms as well, while the pecking order theory predicts the opposite direction since large firms usually have more cash in hand with controlling for investment.

Leverage: higher leverage ratio implies a higher bankruptcy risk that firm will face with, resulting a higher cash level that firm is willing to hold. However, pecking order theory argues a negative relationship between leverage ratio and cash holdings.

Market to book ratio: this variable stand for firms’ investment opportunity, both trade-off and pecking order theory predict a positive correlation. Firms tend to stock extra cash if they have more investment opportunities since the higher cost of external financing. However, free cash flow model argues in a different way, which implies that firm managers need to ensure there is sufficient internal source to invest growth projects when firm has poor investment opportunities at the moment.

Dividend dummy: trade off theory predicts a negative correlation since firms that pay dividend can raise funds at a relatively lower cost in the future by cutting the dividend. While the other two theories fail to provide support of the effect of this variable.

5. Descriptive statistics

Table one reports the details of the variables that are used in this paper. Average cash ratio is 0.262, with 0.258 standard deviations. HHI (Herfindahl-Hirschman Index) is scaled by 10000 for the ease of interpretation. The median of HHI is 0.056 and its minimum and maximum amount are 0.0003 and 0.467 respectively. The range of firm position is between 0 and 1 with 0.103 as the median level. All other control variables’ descriptive statistics are similar to previous literature except for market to book ratio. The average of market to book ratio in this sample is 2.6321, which is relatively higher compared with previous literature

(16)

records. However, the sample is from quite recent years (2002-2016). It could be the firm’s investment opportunities are higher as time going since market to book ratio is a proxy to represent the firm’s investment opportunities.

Table 1

This table is the descriptive statistics of firm-year variables over 2002 to 2016. Cash ratio is calculated as cash and short-term investments divided by assets. Size is defined as the natural logarithm of assets. Leverage ratio is calculated as the sum of long term debt and debt in current liabilities over assets. Market to book ratio is calculated as market value plus book debt and divided by assets. Cash flow volatility is measured as the standard deviation of quarterly cash flow ratio, while the measurement of cash flow to assets follows the ideas from Gormley and Matsa (2016), who used (operating income before depreciation-accruals)/ current

assset[_n-1], where accruals is equal to (current assets-current assets[_n-1])-( cash and short-term

investments - cash and short-term investments [_n-1])-(current liabilities- current liabilities [_n-1])+(debt in current liabilities- debt in current liabilities [_n-1])-depreciation and amortization, where [n_1] stands for the last quarter in the same year. Dividend dummy is a dummy variable equal to one if dividends-common reported in Compustat database. Other variables are research and development amount, net working capital, acquisition, and capital expenditure, the measurements of them are all scaled by assets.

Variable Obs Mean Std.dev Min Median Max

Cash ratio 29549 0.2620 0.2576 0.0000 0.1689 0.9537

Size 29552 5.3624 2.5665 -1.5750 5.3177 11.1769

Leverage 29463 0.2903 0.6015 0.0000 0.1585 5.5074

Market to book ratio 28155 2.6321 4.7979 0.2471 1.4027 43.8144

R&D /assets 24328 0.1508 0.2646 0.0000 0.0622 1.9514

Net working capital/assets 29398 -0.0757 0.9782 -9.4773 0.0671 0.5263

Acquisition/assets 28627 0.0191 0.0562 -0.0066 0.0000 0.3420

Capital expenditure/assets 29523 0.0377 0.0412 0.0000 0.0256 0.2528

Cash flow volatility 25279 0.0733 0.1769 0.0000 0.0310 2.7293

Dividend dummy 29552 0.2731 0.4455 0.0000 0.0000 1.0000

HHI 29552 0.0733 0.0570 0.0003 0.0557 0.4672

Firm position 28506 0.1911 0.2084 0.0000 0.1029 0.9999

Table two is the summary statistics under different competition intensity among the industries. The product market competition has been considered as a tool to adjust firm’s internal mechanism. It would be meaningful to check data characteristics of firms under different competition level, with using the median of HHI as the benchmark. The average of almost all variables is statistically significant under different groups at a reasonable percent level. More specifically, the average cash ratio of firms that are operating in the relatively monopoly industries (HHI > median HHI sample) is statistically higher than the average cash ratio of firms which are in the relatively competitive environment. The same pattern shows for firm size, leverage, and market to book ratio, research and development ratio and

(17)

divided dummy. The difference of cash flow volatility under different groups is not statistically significant.

Table 2

The first column reports the mean and standard deviation (in the parenthesis) of the main financial data of the firms that are operating in the relatively competitive industry (Herfindahl-Hirschman Index < median sample) over 2002-2016. Column two represents financial statistics of the firm that operates in the relatively monopoly industry (Herfindahl-Hirschman Index > median sample). The third column shows the significant level of the difference between two different groups of firms.

hhi<median hhi>median p-value of difference

(1) (2) (3) Cash ratio 0.2487 (0.2427) (0.2710) 0.2753 0.0000 Size 5.1098 (2.4094) (2.6910) 5.6150 0.0000 Leverage 0.2811 (0.5902) (0.6126) 0.2996 0.0082

Market to book ratio 2.5673

(4.4365) (5.1366) 2.6976 0.0228

R&D /assets 0.1350

(0.2448) (0.2826) 0.1670 0.0000

Net working capital/assets -0.0497

(0.9374) (1.0169) -0.1018 0.0000

Acquisition/assets 0.0212

(0.0592) (0.0528) 0.0169 0.0000

Capital expenditure/assets 0.0357

(0.0394) (0.0428) 0.0396 0.0000

Cash flow volatility 0.0735

(0.1791) (0.1746) 0.0731 0.8735

Dividend dummy 0.2569

(0.4370) (0.4534) 0.2892 0.0000

Instead of plotting firms’ cash ratio over time with respect to two different HHI levels (use the median or mean of the whole sample’s HHI as the benchmark for example), like Morellec, Nikolov and Zucchi (2009) did. Figure one plots average cash ratio of firm-year over different quantiles of HHI level. Twenty groups are created by 5 percent quantile of HHI each. The graph roughly confirmed the first hypothesis, the correlation between

competition intensity and cash holding is not linear based on the below figure. It shows an increase-decrease trend. Additionally, firms' cash holding level fluctuates a lot among firms that are operating in the relatively monopoly environment. Firms that are operating in the industries which have the median level of HHI have the highest average cash ratio. However,

(18)

no accurate conclusions could be obtained through the graph since HHI is only divided into 20 groups by 5% quantile each. It could only give us the rough idea. The more sensitive analysis is showing in the following session.

Figure 1

The horizontal axis stands for the quantile range of HHI, for example, 5%-10% represents the firms that operate in the industry where competition intensity is in the range of 5% and 10% quantile of HHI. The vertical axis represents the average cash ratio among all firms, all years within the corresponding industries.

6. Empirical results

This paper first concentrates on the firms that are operating in the very competitive environment and test the second hypothesis, which argues that cash ratio increases with competition level if firms are in the very competitive industries. Instead of using a specific benchmark of HHI to represent competition level (previous literature usually choose HHI<0.01 stands for the very competitive environment and HHI>0.18 represents the very monopoly industries), this paper considers the range of HHI of all firms in the manufacturing sector and use 10 percent quantile as the measurement. More specifically, for firms whose HHI is smaller than 10% quantile of HHI of the whole sample is allocated to very competitive environment group. Table 3 shows the empirical results, model one and two focus on the inter-competition, whereas model three and four consider intra-competition as well. All models are using pooled OLS instead of fixed effect model since industry concentration variable only change every five years. Due to the research question of this paper, industry effect cannot be controlled either. Therefore, all models are using simple OLS model, with

0 0.1 0.2 0.3 0.4 0.5 0.6 av er ag e c as h r at io

quantile range of HHI

(19)

White’s correction for heteroscedasticity consideration. All four models confirm the second hypothesis, the negative relationship between HHI and cash ratio stands for a positive correlation between competition intensity and cash holing level since higher HHI represents weaker competition. Predatory behavior is most unlikely to happen in the very competitive industries and firm stock extra cash due to high competitive pressure and higher marginal benefits of holding extra cash in the competitive environment. There are not too many differences with or without controlling time specific factors. After controlling firm position, which is negatively correlated with firm cash holding level, the positive relationship between competition and cash holding becomes more severe, the coefficient before HHI change from -1.23 to -1.90, the significance level changes from ten percent to one percent. Even the negative coefficient before firm potion variable confirms the fourth hypothesis, the effect is not significant at all, the intuition of this result is that firms are already very similar to their rivals within the corresponding industry due to the fact that they are in the very competitive industries and their position wouldn’t be very different. Therefore, the effect of firm

position does not matter that much. All control variables show the same directions with previous literature (Opler, Pinkowite, Stulz and Williamson(1999) and Haushalter, Klasa and Maxwell(2007)), and all control variables are statistically significant except net working capital and cash flow volatility.

(20)

Table 3

This table presents the analysis of the manufacturing firms that are operating in the very competitive environment, where Herfindahl-Hirschman Index is smaller than 10 percent quantile of HHI, over 2002-2016 period. The dependent variable is cash ratio, which defined as the sum of cash and short-term investments divided by assets. All four models are pooled OLS regressions with White’s correction for heteroscedasticity corrected purpose. *** ** * stand for 1% 5% and 10% significant level and standard errors are showed in the parentheses. Model one and three are not controlling time effect, whereas model two and four consider cash ratio might change over time with controlling time effect.

HHI<10% quantile of HHI (very competitive environment)

（_1）（_2）（_3）（_4） HHI -1.2319* (0.6374) -1.2499* (0.6391) -1.9030*** (0.6352) -1.9063*** (0.6376) Firm position -0.0192 (0.0219) -0.0211 (0.0219) Size -0.0079*** (0.0023) -0.0086*** (0.0023) -0.0078*** (0.0024) -0.0083*** (0.0024) Leverage -0.1121*** (0.0160) -0.1107*** (0.0159) -0.1075*** (0.01546) -0.1066*** (0.0155) Market to book ratio 0.0057**

(0.0024) 0.0055** (0.0024) 0.0059** (0.0024) 0.0058** (0.0025) R&D /assets 0.1976*** (0.0613) 0.1955*** (0.0607) 0.2120*** (0.0628) 0.2102*** (0.0624) Net working capital/assets -0.0089

(0.0092) -0.0070 (0.0092) -0.0077 (0.0091) -0.0063 (0.0091) Acquisition/assets -0.3434*** (0.04167) -0.3444*** (0.0432) -0.3273*** (0.04197) -0.3276*** (0.0435) Capital expenditure/assets -0.4107*** (0.0865) -0.4106*** (0.0864) -0.3306*** (0.0883) -0.3272*** (0.0882) Cash flow volatility 0.01795

(0.0354) 0.0168 (0.0352) 0.0189 (0.0356) 0.0185 (0.0355) Dividend dummy -0.0254*** 0.0075 -0.0261*** (0.0075) -0.0203*** (0.0077) -0.0210*** (0.0077)

Time fixed effect No Yes No Yes

Obs 1815 1815 1744 1744

Adjust R^2 0.1875 0.1916 0.1923 0.1954

(21)

Table 4 aims to investigate the effect of competition and firm position on cash ratio among firms that are in another extreme case, which is very monopoly environment. This paper considers the HHI of the corresponding industry is higher than 90% quantile of the whole sample HHI as the very monopoly case. All four models confirm the third hypothesis. For firms that operate in the very concentrated industries, higher monopoly, higher cash ratio. For the first two models, which consider inter-competition alone, one unit increment in HHI results in 0.2549 higher of cash ratio level, the effect is statistically significant at one percent level. If firms are in the very monopoly industries, a higher level of HHI stands for a higher possibility of the occurrence of predatory behavior, compared to the positive effect of competition on cash ratio, which is showed in the previous table, firms concern more about the predatory risk. Therefore, the negative effect of competition on cash ratio dominants in the very concentrated environment group. Model three and four control firm position effect as well, the coefficient is around -0.1, which is much larger than the coefficients in table3(-0.02) and its statistically significant. This result confirms the last two hypotheses. firms which have median productivity level within the industry tend to hold more cash because the more similarity investment opportunities that they meet with their rivals, the higher predatory risk. Intra-competition mattes more if firms are in the monopoly environment since this specific environment is the precondition of the occurrence of predatory behavior, and the shorter distance that firm away from the middle production level of the

corresponding industry, the greater the predatory risk, resulting in a higher level of cash holdings. The effects of other control variables keep the same with previous table, except for cash flow volatility. However, this effect is not statistically significant.

(22)

Table 4

The subsample in this table are the manufacturing firms that are operating in the very monopoly environment, where Herfindahl-Hirschman Index is higher than 90 percent quantile of HHI, over 2002-2016. All four models are pooled OLS regressions that regress on the sum of cash and short-term investments divided by assets with White's correction for heteroscedasticity corrected purpose. *** ** * stand for 1% 5% and 10% significant level and standard errors are showed in the parentheses. Only model two and model four considers the time specific factors.

HHI>90% quantile of HHI (very monopoly environment)

（1）（2）（3）（4） HHI 0.2549*** (0.0742) 0.2414*** 0.0753 0.2855*** (0.0782) 0.2743*** (0.0796) Firm position -0.1090*** (0.0211) -0.1055*** (0.0211) Size 0.00012 (0.0019) -0.0005 (0.0020) -0.0012 (0.0020) -0.0018 (0.0020) Leverage -0.1706*** (0.0184) -0.1720*** (0.0186) -0.1683*** (0.0184) -0.1699*** (0.0186) Market to book ratio 0.0072***

(0.0022) 0.0075*** (0.0023) 0.0082*** (0.0022) 0.0084*** (0.0023) R&D /assets 0.2370*** (0.0426) 0.2332*** (0.0427) 0.2039*** (0.0408) 0.1997*** (0.0409) Net working capital/assets -0.0229

(0.0122) -0.0231* (0.0122) -0.0246** (0.0119) -0.0252** (0.0120) Acquisition/assets -0.30625*** (0.0589) -0.3083*** (0.0609) -0.3178*** (0.0609) -0.3214*** (0.0629) Capital expenditure/assets -0.5798*** (0.1139) -0.5494*** (0.1169) -0.5935*** (0.1199) -0.5608*** (0.1230) Cash flow volatility -0.01415

(0.0735) -0.0204 (0.0734) 0.0004 (0.0718) -0.0071 (0.0721) Dividend dummy -0.1155*** (0.0085) -0.1162*** (0.0085) 0.0718*** (0.0088) -.1133*** (0.0089)

Time fixed effect No Yes No Yes

Obs 1607 1607 1528 1528

Adjust R^2 0.3302 0.3380 0.3493 0.3579

(23)

After focusing on the two extreme cases, it would be meaningful to investigate the correlation between industry concentration and cash ratio among firms that operating in the rest of the sample in order to see the whole picture of competition. The results are showed in table 5. The first two models show the corresponding results. It shows that one unit increase in HHI results in 0.49 unit increase in cash ratio, which means that cash ratio increase as the competition becomes less among firms that operating in the relatively normal environment (neither very competitive nor very monopoly). The result implies that the negative effect of competition on cash ratio because of the predatory behavior

dominants in the more general case and this negative effect is statistically significant at one percent level. With respect to the effect of firm position, the coefficient is -0.0418 at any reasonable significance level. and it is in the range between -0.02(the coefficient of firm position in the very competitive group) and -0.1(the coefficient of firm position in the very monopoly group). This results confirm the fifth hypothesis to some extent, the effect of firm position becomes more important as the monopoly level increases.

Another subsample (45% – 55% quantile of HHI) in table 5 is chosen for the following considerations. Firstly, firms that operating in the industries which HHI around the median of HHI of the whole sample has the highest average cash ratio according to the figure 1. Secondly, it would be interesting to investigate how would competition intensity affect cash ratio if firms are operating in the industries that have the median HHI of the whole

manufacturing sector. Therefore, 45%-55% quantile of HHI is chosen as a specific group to test the relationship between competition and cash ratio. The results are shown in model 3 and model 4 in table 5. The result is quite surprising. The coefficient before HHI becomes very large, which is around 12.7, and this effect is statistically significant at one percent level. The coefficient is almost the same with and without controlling firm position factor.

Moreover, firm position is not statistically significant to influence cash ratio for firms that operating in the environment which has median HHI. The possible explanation for this is firms that are in the median HHI industries are experiencing the transform stage, whether the corresponding industry is going to become relatively competitive or relatively monopoly. All firms in this kind of environment will be more willing to implement predatory behavior not only to their rivals that already exist in the corresponding industry, but also prevent the future entrants to entry into the industry. And the more monopoly the corresponding

(24)

industry in this specific group is, the higher possibility that the HHI of the industry will be higher than the median of whole manufacturing sector's HHI level in the near future, thus firms are willing to stock more cash for predatory behavior. Compared with the remarkable effect of inter-competition on cash ratio in this specific group, the influence of

intra-competition (firm position) is not obvious. Even the coefficient is still negative, the effect is not significant any more, which implies that firms that operate in this group concern more about the potential entrants since firm position fails to explain the higher average cash ratio in this group. Additionally, all other control variables have the same pattern with previous analysis except firm size. It shows a positive relationship with cash ratio at one percent significant level. With respect of the effect of size on firm cash holdings, trade-off theory predicts a negative correlation because of the economies of scales in liquid assets. The significantly positive relationship that showed in model 3 and 4 in Table 5 relatively confirm the explanation before, firms that in this kind of environment are willing to become more monopoly in the future, thus large firms tend to hold more cash since they are more able to conduct predatory behavior to both existing and potential rivals.

(25)

Table 5

There are two subsamples in this table. The first group contains the manufacturing firms that are operating in neither very competitive nor very monopoly industry, where Herfindahl-Hirschman Index is between 10 percent and 90 percent quantile of HHI, over 2002-2016. The other one is the firms that have median HHI, positions in 45%-55% quantile of whole sample HHI, over 2002-2016. All four models are pooled OLS regressions with White’s correction for heteroscedasticity corrected purpose. *** ** * stand for 1% 5% and 10% significant level and standard errors are showed in the parentheses.

10% – 90% quantile of HHI 45% – 55% quantile of HHI

（1）（2）（3）（4） HHI 0.4875*** (0.0526) 0.4812*** (0.0534) 12.6713*** (2.4204) 12.6675*** 2.4547 Firm position -0.0418*** (0.0098) -0.01849 (0.0277) Size 0.0004 (0.0009) 0.0003 (0.0009) 0.01292*** (0.0027) 0.0129*** (0.0027) Leverage -0.1734*** (0.0088) -0.1715*** (0.0088) -0.2164*** (0.0265) -0.2144*** (0.0263) Market to book ratio 0.0050***

(0.0008) 0.0048*** (0.0008) 0.0064*** (0.0019) 0.0062*** (0.0019) R&D /assets 0.3428*** 0.0122 .3386*** (0.0122) 0.3313*** (0.0216) 0.3291*** (0.0215) Net working capital/assets -0.0257***

(0.0061) -.0250*** (0.0063) -0.0422** (0.0177) -0.04141** (0.0175) Acquisition/assets -0.6669*** (0.0198) -.6802*** (0.0201) -0.8974*** (0.0586) -0.9153*** (0.0598) Capital expenditure/assets -1.3891*** (0.0464) -1.3928*** (0.0467) -1.8929*** (0.1243) -1.9182*** (0.1253) Cash flow volatility -.0283***

(0.0203) -.0209 (0.0204) -0.0580 (0.0479) -.0539 .0477 Dividend dummy -0.13889*** (0.0039) -.01381*** (0.0040) -0.1237*** (0.0118) -0.1212*** (0.0119)

Time fixed effect Yes Yes Yes Yes

Obs 16404 16041 2,865 2834

Adjust R^2 0.3332 0.3304 0.3844 0.3797

(26)

7. Robustness check

To ensure the above results are indeed reliable, two robustness checks are conducted in this section. Only the models that considered both industry concentration and firm position, with controlling time specific factors are chosen to conduct the robustness check.

The first robustness check considers the different measurement of cash holding level. There are three famous measures, cash/assets, cash/net assets and the natural logarithm of cash/net assets respectively. This paper uses the first one, which is the most traditional one as the proxy to represent firm cash ratio level. The other two measures would have very serious outliners’ problem. This paper uses the second measurement, which is cash/net assets as the dependent variable to recheck the correlation between cash ratio and competition. However, due to the extreme outliners, this variable is winsorized at 15 percent level. Table 6 presents the results of the new regressions which use cash/net assets as the dependent variable. The results do not have too many differences, the direction and significance level of the two main variables are consistent with the previous results. The second robustness check considers the measurement of the size of firms. This paper uses the natural logarithm of total assets as the proxy to represent firm size. This robustness section chooses logarithm of sales as another way to stand for firm size. The results are very similar to precious analysis and confirm the previous finding based on Table 7. However, there is one change that worth to mention, the coefficient before size become significant negatively correlated with cash holdings for all four models.

(27)

Table 6

This robustness check section contains four groups that have dividend by the whole sample. These groups contain the manufacturing firms that are operating in the environment which HHI is smaller than 10% quantile, between 10%-90% quantile, larger than 90% quantile, between 45%-55% quantile of HHI whole sample respectively, over 2002-2016 time period. The table shows the results of regressions models with controlling time-specific factors, using cash/net assets as the dependent variable. All models are pooled OLS regressions with White’s correction for heteroscedasticity corrected purpose. *** ** * stand for 1% 5% and 10% significant level and standard errors are showed in the parentheses.

Dependent variable: cash/net assets (with 15% winsorized) HHI <10% quantile of HHI (1) HHI 10% – 90% quantile of HHI (2) HHI >90% quantile of HHI (3) HHI 45% – 55% quantile of HHI (4) HHI -3.6495*** (1.5697) 0.9497*** (0.1037) 0.6205*** (0.1646) 20.1723*** (4.4625) Firm position -0.0290 (0.0400) -0.0759*** (0.0178) -0.1853*** (0.0434) -0.0296 (0.0475) Size -.0172*** (0.0045) -0.0025 (0.0018) -0.0067* (0.0040) 0.0182*** (0.0051) Leverage -0.1868*** (0.0277) -0.2945*** (0.0158) -0.3262*** (0.03644) -0.3522*** (0.0421) Market to book ratio 0.0105**

(0.0047) 0.0094*** (0.0015) 0.0153*** (0.0044) 0.0118*** (0.0034) R&D /assets 0.4208*** (0.1197) 0.5785*** (0.0214) 0.4085*** (0.0864) 0.5533*** (0.0374) Net working capital/assets -0.0085

(0.0170) -0.0411*** (0.0112) -0.0544** (0.0233) -0.0664* (0.0283) Acquisition/assets -0.6085*** (0.0760) -1.2182*** (0.0374) -0.5872*** (0.1232) -1.5841*** (0.1104) Capital expenditure/assets -0.6870*** (0.1615) -2.4187*** (0.0853) -1.1503*** (0.2493) -3.1932*** (0.2298) Cash flow volatility 0.0346

(0.0670) -0.0723 (0.0346) -0.0801 (0.1359) -0.1501 (0.0789) Dividend dummy -0.0421*** (0.01456) -0.2512*** (0.0076) -0.2080*** (0.0184) -0.2039*** (0.0236)

Obs 1744 16041 1528 2834

Adjust R^2 0.1947 0.3133 0.3244 0.3490

(28)

Table 7

This robustness check section contains four groups that have dividend by the whole sample. These groups contain the manufacturing firms that are operating in the environment which HHI is smaller than 10% quantile, between 10%-90% quantile, larger than 90% quantile, between 45%-55% quantile of HHI whole sample respectively, over 2002-2016 time period. The table shows the results of regressions models with controlling time specific factors, using cash/assets as the independent variable. All models are pooled OLS regressions with White’s correction for heteroscedasticity corrected purpose. *** ** * stand for 1% 5% and 10% significant level and standard errors are showed in the parentheses. All other control variables are the same as before except for size only, the size of the firm is constructed as the logarithm of sales instead.

Dependent variable: cash/ assets HHI <10% quantile of HHI (1) HHI 10% – 90% quantile of HHI (2) HHI >90% quantile of HHI (3) HHI 45% – 55% quantile of HHI (4) HHI -1.7711*** (0.6295) 0.6804*** (0.0523) 0.2865*** (0.0790) 6.4955*** (2.4497) Firm position -0.0367* (0.0211) -0.05814*** (0.0097) -0.1116*** (0.0208) -0.0226 (0.0277) Size -0.0187*** (0.0026) -0.0272*** (0.0009) -0.0063*** (0.0022) -0.0269*** (0.0021) Leverage -0.1027*** (0.0144) -0.1565*** (0.0080) -0.1688*** (0.0185) -0.1860*** (0.0221) Market to book ratio 0.0047**

(0.0024) 0.0029*** (0.0008) 0.0079*** (0.0023) 0.0044** (0.0019) R&D /assets 0.1792*** (0.0567) 0.2620*** (0.0120) 0.1893*** (0.0400) 0.2408*** (0.0217) Net working capital/assets -0.0008

(0.0085) -0.0179*** (0.0058) -0.0259** (0.0119) -0.0236 (0.0148) Acquisition/assets -0.2761*** (0.0420) -0.5179*** (0.0189) -0.3122*** (0.0627) -0.6640*** (0.0535) Capital expenditure/assets -0.2821*** (0.0843) -1.1689*** (0.0436) -0.5412*** (0.1234) -1.5752*** (0.1143) Cash flow volatility -0.0142

(0.0339) -0.0920 (0.0197) -0.0330 (0.0760) -0.1310 (0.0471) Dividend dummy 0.0006 (0.0073) -0.0625*** (0.0037) -0.1022*** (0.0089) -0.0194* (0.0107)

Obs 1744 16041 1528 2834

Adjust R^2 0.2284 0.3736 0.3621 0.4106

(29)

8. Conclusion and limitation

This paper uses all publicly traded firms that are in the manufacturing sector in the US over 2002-2016 to investigate the effect of product market competition on firms’ cash holding level. Industry concentration and firm position represent for inter competition and intra competition respectively. Instead of considering the sample as a whole, several groups are divided based on industry concentration, very competitive environment (< 10% quantile of HHI), normal environment (varies between 10% and 90% quantile of HHI), very monopoly environment (>90% quantile of HHI), and the median group (45% - 55% quantile of HHI). The first finding of this paper is the non-linear relationship between industry concentration and cash holdings, the effect of industry concentration level on cash holding depends on which group the firm is in. If the firm is operating in a very competitive environment, firm cash stock level has an upward trend with the higher competition intensity of the

corresponding industry. The correlation between industry concentration and firm cash holdings has an opposite pattern if the firm is in the normal and very monopoly

environment. The above effects remain the same direction after controlling firm position, which represents the similarity among firms within the industry. Unlike the

non-monotonous effect of inter competition on cash holdings, the more similarity of the rest of firms within the industry, the more cash that firms hold among all groups. However, the influence of firm position is not significant if firms are in the very competitive environment and another specific group (45 percent to 55 percent quantile of HHI of the whole sample). The effect of firm position is the most important if firms operate in the very monopoly industry, where the predatory behavior is most likely to happen. All the conclusions are reliable since the results are quite similar after robustness checks which are conducted in section 7. Additionally, the effect of control variables which represent firm's own

characteristic is quite similar with the findings obtained by Opler, Pinkowitz, Stulz, and Williamson (1999), thus adding product market competition analysis is not conflict with the traditional explanation of the determinants of cash holding.

The results of this paper advise firms’ manager to consider both industry concentration and its own position within the industry when taking actions for product market competition. It wouldn’t be wise to conduct predatory behavior if firms are already in the very competitive

(30)

environment. And if firms are in the core of the corresponding industry, where is a quite concentrate environment, firms need to be more aware of the predatory behavior. Even all five hypotheses that mentioned before is proven empirically using all publicly traded manufacturing firms in the US over 2002-2016, this paper does have some limitation. Firstly, like all other empirical methods, the results rely heavily on the data itself, and some groups (10% quantile and 90% quantile of HHI) only have around 1500 observations, which could possibly make the results only suit this specific sample over this specific time period. Secondly, this paper mainly focuses the effect of product market competition on firm cash holding level, without paying too much attention to the factors such as the ownership of the company, credit rating of the company, etc. even the variables that stand for firms’ crucial characteristic are controlled. Future research is needed to control more about firms’ status and use more sample and longer duration of time period to confirm the results of this paper.

(31)

References

Alimov, A., 2014, Product market competition and the value of corporate cash, Journal of Corporate Finance 25, 122-139.

Bates, T.W., Kahle, K.M., Stulz, R.M., 2009, Why do U.S. firms hold so much more cash than they used to. The journal of finance, Vol.LXIV,No.5.

Bolton, P., Scharfstein, D., 1990, A Theory of Predation Based on Agency Problems in Financial Contracting, American Economic Review 80, 93-106.

Chou, J., Ng, L., Sibilkov, V., Wang, Q., 2011, Product market competition and corporate governance, Review of Development Finance 1, 114-130.

Denis, D., Sibilkov, V., 2012, Financial constraints, investment, and the value of cash holdings, Review of Financial Studies 23, 247-269.

Fresard. L., 2010, Financial strength and product market behavior: the real effects of corporate cash holdings, Journal of Finance 65, 1097-1122.

Gormley, T.A., Matsa. D.A., 2016, Playing it safe? Managerial preferences, risk and agency conflicts, Journal of Financial Economics 122, 431-455.

Grill, A., Mathur, N., 2011, Factors that influence corporate liquidity holdings in Canadas, Journal of Applied Finance and Banking 1, 133-153.

Haushalter, D., Klasa, S., Maxwell, W.F., 2007, The influence of product market dynamics on a firm’s cash holdings and hedging behavior, Journal of Financial Economics 84, 797-825. Islam, S., 2012, Manufacturing firms cash holding determinants: Evidence from Bangladesh, International Journal of Business and Management 7, 172-184.

Kim, C., Mauer,D., Sherman,D., 1998, The determinants of corporate liquidity: Theory and evidence, Journal of Financial and Quantitative Analysis 33, 335-359.

Lyandres, E., Palazzo, B., 2016, Cash holding, competition and innovation, Journal of Financial and Quantitative analysis 51, 1823-1861.

(32)

Mackay, P., Phillips, G.M., 2005, How does industry affect firm financial structure? Review of Financial Studies 18, 1433-1466.

Mikkelson, W.H., Partch,M., 2003, Do persistent large cash reserves hinder performance? Journal of Financial and Quantitative Analysis 38, 275-294.

Morellec, E., Nikolov, B., Zucchi, F., 2009, Competition, cash holdings, and financing decisions, Swiss Finance Institute Research Paper, No.13-72.

Nickell, S.J., 1996, Competition and corporate performance, Journal of Political Economy 104, 724-746.

Opler,T., Pinkowitz, l., Stulz, R., Williamson, R., 1999, The determinants and implications of corporate cash holdings, Journal of Financial Economics 52, 3-46.

Schmidt, K., 1997, Managerial Incentives and Product Market Competition, Review of Economic Studies 64, 191-213.

Siems, M., 2008, Shareholder protection around the world, Delaware Journal of Corporate Law 33.

Yang, X., Zheng, J., 2017, The impact of predation risk on the marginal value of cash holdings: An empirical perspective, Journal of Financial Risk Management 6, 150-162.

The effect of product market competition on cash holdings