An empirical analysis on whether Chinese state-owned firms are tended to use derivatives speculatively

(1)

An Empirical Analysis on Whether Chinese

State-Owned Firms are Tended to Use Derivatives

Speculatively

14th December, 2015

Master thesis

Author: Yingjie Tan (10758143) Supervisor: Dr. J.E. Ligterink

Study program: MSc Business Economics, Finance track

(2)

Author’s Declaration of Originality

I hereby certify that I am the sole author of this thesis and that no part of this thesis has been published or submitted for publication.

I certify that, to the best of my knowledge, my thesis does not infringe upon anyone’s copyright nor violate any proprietary rights and that any ideas, techniques, quotations, or any other material from the work of other people included in my thesis, published or otherwise, are fully acknowledged in accordance with the standard referencing practices. Furthermore, to the extent that I have included copyrighted material that surpasses the bounds of fair dealing within the meaning of the Canada Copyright Act, I certify that I have obtained a written permission from the copyright owner(s) to include such material(s) in my thesis and have included copies of such copyright clearances to my appendix.

I declare that this is a true copy of my thesis, including any final revisions, as approved by my thesis committee and the Graduate Studies office, and that this thesis has not been submitted for a higher degree to any other University or Institution.

(3)

Abstract

In this study, we test whether state-owned companies have less incentive to use derivatives to reduce risk and are more tended to use derivatives speculatively. This paper examines the relationship between the probabilities of derivative-use and company’s ownership by establishing logistic regression and controlling other firm characteristics. The sample is 340 list companies in manufacturing industry. The research window is year 2004. But no clear relationship is found. We also study the interaction effect of derivative-use and company’s ownership on risk by using OLS regression. The effect is significantly negative on systematic risk, which indicates that state-owned firms are hedging against systematic risk.

(4)

Derivatives have been becoming one of the most important financial instruments, as indispensable risk management tools in worldwide top firms. Emerging and development of derivative markets is one of the most prominent changes in international financial markets in recent 40 years. The rapid spread and extensive use of derivatives moved financial innovation, which arose in 1970s, into a climax and deeply affected the whole financial markets (Hull, 2006). The main reason is that the market provides an effective tool monitor financial risks at relatively low cost for enterprises and financial institutions. Derivatives are originally created for price discovering, hedging and transferring risks (Zhao, 2011).

At its 24th Annual General Meeting in Beijing, the International Swaps and Derivatives Association, Inc. (ISDA) announced the results of a survey of derivatives usage by the world's 500 largest companies. According to the survey, 94% of these companies use derivative instruments to manage and hedge their business and financial risks (2009 ISDA Operations Benchmarking Survey).

This is the second such survey conducted by ISDA, and the first was in 2003. The results of this most recent survey show that derivatives use among large corporations continues to grow. The companies included in the survey are headquartered in 32 different countries and represent a broad range of industries from basic materials to office equipment to retail and even health care. The survey found that the use of derivatives is common to companies worldwide: among the ten countries with the largest number of the 500 companies surveyed, all companies based in Canada, France, Great Britain, Japan and The Netherlands report using derivatives while 97 percent of German companies and 92 percent of US companies report using derivatives. Companies in South Korea and China were least likely to report using derivatives, but

(6)

87 percent of Korean companies and 62 percent of Chinese companies nonetheless do report using these instruments (2009 ISDA Operations Benchmarking Survey).

Figure 1. Derivative usage by Global Fortune 500

Figure 2. Top 10 countries with highest reported use of derivatives

From this perspective, derivatives are generally accepted by world top companies as necessary risk management tools.

Chinese economy is an emerging market. After joining the World Trade Organization, domestic Chinese market is related to international market more and more closely, which means prices in two markets have mutual effects on each other. The costs,

(7)

commodity prices and financial situation of Chinese export enterprises will be affected by foreign exchange volatility. Considering the huge risks brought to enterprises by fluctuating prices, Chinese government started to permit some large-scale state-owned companies to hedge through derivatives traded overseas. Meanwhile, the government proposed to domestic futures exchanges, such as Shanghai Futures Exchange, to increase transaction variety of futures, which gives Chinese enterprises more choices to hedge risks (Zhao, 2011).

State-owned Assets Supervision and Administration Commission of the State Council (SASAC) insists that Chinese state-owned firms should stick to hedging, because derivative transactions require low up-front fee and the value of derivatives is highly volatile (Notification of Furthering the Supervision of State-owned Enterprises’ Financial Derivative Transactions 2009). However, many Chinese state-owned firms have reported gigantic losses in using derivatives to hedge risks. Air China and China Eastern Airlines announced a 310 billion RMB and a 183 billion RMB loss of fair value of hedging instruments on fuel before Oct. 31, 2008 respectively (CFFEX Institute for Financial Derivatives).

According to hedging theory, spot and future prices of the same underlying assets almost synchronize and the gains or losses in sport markets can be partly or even perfectly offset by gains or losses in derivative markets (Hull, 2006). Forward contracts are designed to neutralize risk by fixing the price that the hedger will pay or receive for the underlying asset. Option contracts, by contrast, provide insurance. They offer a way for investors to protect themselves against adverse price movements in the future while still allowing them to benefit from favorable price movements (Hull, 2006). Companies should gain or lose less when taking hedging strategies than only investing in spot markets. It is unusual to suffer huge losses when firms are motivated to hedge. After discussing the principal-agent problem in Chinese state-owned companies and CITIC Pacific’s usage of Knock-Out Discount Accumulator (KODA), analysts argued these companies were speculating, which means taking more risks, instead of hedging risks

(8)

as they declared in their bulletins (Li, and Liao, 2011).

1.2 Research question

After closely examining the ownership structure of some companies which reported big losses, we find that the government is the biggest shareholder at around 40%, which may invites another “tragedy of the commons”. Take Air China as example (shown in the following chart). China National Aviation Holding Company and China National Aviation Corporation (Group) Limited, both of which are state-owned, hold 41% and 12% shares of the company.

Figure 3. Ownership structure of Air China

Imagine assets with normal distributed returns (see Figure 4), the safer the assets, the bigger the probability of negative returns (fatter tail). If returns of assets tend to be very close to the mean (expected value) and hence to each other, it is unlikely we will lose much. And if returns of assets are very spread out around the mean and from each other, it is more likely the loss will be tremendous.

(9)

Figure 4. Distributions of asset with different levels of risk

For state-owned companies, the executives have incentives to take more risk at the cost of taxpayers: if they win the gamble, they can harvest the fat profit as Air China and China Eastern Airline state in their 2008 annual reports that payment for directors, supervisions and senior managers is linked to company’s profits (Li, and Liao, 2011); if they lose, the bailout (also known as subsidies to loss-making companies) will be paid by the government, or more specifically taxpayers. Then state-owned companies are expected to undertake more risky investment and become riskier and more volatile. We can study the volatility of stock return and research whether state-owned derivative users are more volatile than non-state-owned derivative users.

In countries implementing International Accounting Standards (IAS) and those with mature derivative markets, derivative using companies are ruled by a relatively well-developed accounting system and supervised more strictly by regulators. However, studies on Chinese accounting standards of derivatives started in 1990s and there is still much to improve (Fu, 2001). A lack of powerful regulation on disclosure of financial information of derivatives also potentially encourage companies to take more risks. So the research question for is as follows:

“Do Chinese state-owned companies use derivatives to speculate rather than to hedge risks?”

First, I will study whether state-owned firms have lower probability of using derivatives. Second, I will examine whether derivatives used by state-owned enterprises will

(10)

produce more risks than non-state firms.

1.3 Contribution

First, there are very limited studies focusing on the relationship between derivatives usage and state-owned companies’ investment behaviors. Most of them focus on analyzing specific derivative trading strategy in a specific company. Li and Liao (2011) study on the derivative contracts of CITIC Pacific, Air China, China Eastern and China COSCO in year 2008, in which these state-owned firms lost gigantically in derivatives. But they did not perform any empirical studies.

In our research, econometrical analyses, like logistic and OLS regression, are conducted to examine state-owned firms in the manufacturing industry for a whole picture of derivatives usage and risk.

Although annual reports reveals whether a company use derivatives during reporting years, there are few listed companies disclose characteristics of their contracts, like underlying assets (interest rate, foreign exchange, or commodity) and notional principal values of derivative holdings. Hentschel and Kothari (2001) split the sample into three subgroups by the underlying assets of derivatives and the ratio of nominal contract value to total assets. Even though those more precise factors cannot be included in our research, dummy variable of whether use derivatives are available, which still makes empirical studies possible.

Second, endogeneity problem are not avoided or solved in many papers. The endogeneity problem in the research of Li and Xiao (2010) is excluding that riskier firms will use derivatives to decrease risk exposure when studying the effect of derivative-use on company’s risks. It happens to be the same simultaneity problem that will be discussed in Section 7. We use instrumental variable (IV) method to solve it.

IV method is rarely used because it is hard to find a perfect instrument. Hentschel and 10

(11)

Kothari (2001) seek instruments that are highly correlated with the levels of the endogenous variables, but uncorrelated with their variation due to endogenous considerations, but find no suitable instruments existing. So they create instruments based on the ratio of notional value to total risk, which is not accessible for Chinese sample.

In our research, institutional ownership, which is correlated to systematic risk beta and independent of a firm’s decision to use derivatives for risk management, is employed as IV. This ready index saves the trouble of creating a new one.

1.4 Structure

Section 2 discusses the literature review and important concepts. Section 3 constructs two hypotheses of my study. Section 4 displays data collection and gives a rough statistics summary of data. Section 5 describes the research methodology. Section 6 discusses the results of the empirical test. Section 7 uses another model to give additional test. Section 8 presents the conclusion of the study.

(12)

Section 2: Literature review

In this chapter the most important literature with regard to derivatives usage and risk management is discussed.

2.1 Motivations of employing derivatives

A derivative is defined as a contract whose value directly depends on one or multiple underlying securities, stock index, debt instruments, commodities, other derivatives, or any other arranged price index and agreement; derivatives is a deal based on rights and obligations of underlying assets, but it does not necessarily involve transfer of underlying assets (U.S. Commodity Future Trading Commission Derivative Markets Report 1993).

A large collection of literature explores whether financial risk management can benefit companies by contributing more firm values, which shed some light on motivations of companies applying risk management strategies.

Carter, Rogers and Simkins (2003) investigate the fuel hedging behavior of firms in the US airline industry during 1994-2000 to examine whether such hedging is a source of value for these companies. They found that jet fuel hedging is positively related to airline firm value. However, Koski and Pontiff (1999) investigate investment managers' use of derivatives by comparing return distributions for equity mutual funds that use and do not use derivatives and found that derivative users have risk exposure and return performance that are similar to nonusers. Jin and Jorion (2006) found that hedging did reduce stock price volatility for oil and gas producers, but did not find an increase in firm value.

Modigliani and Miller (1958) state if financial market is perfect, hedging cannot increase firm value without agency cost, information asymmetry, tax and transaction cost. A lot of subsequent researches relax the assumptions of MM theory and justify company’s needs for hedging in theory. Many studies prove that hedging can decrease

(13)

financial distress cost, expected tax liabilities, investment insufficiency and coordinate company’s interested parties. There are four theories as follows:

2.1.1 Reduce financial distress cost

Smith and Stulz (1985) conclude that hedging can decrease volatility of company earnings (cash flow) and the likelihood of financial distress.

Stulz (1996) argues that the primary goal of risk management is to eliminate the probability of costly lower-tail outcomes—those that would cause financial distress costs or make a company unable to carry out its original investment strategy.

Haushalter (2000), Graham and Rogers (2000) find a positive linear relationship of debt and hedging, which is consistent with theoretical prediction of company hedging to reduce expected financial distress cost.

2.1.2 Tax benefits

Smith and Stulz (1985) found when facing progressive tax rates or convex tax function, a company can decrease the volatility of income and then minimize taxes by hedging.

In addition, if interest expenses are tax deductible, the reduction of cash flow volatility can also increase corporate debt capacity and tax benefits (Stulz, 1996; Leland,1998).

Graham and Rogers (2002) finds a company will not hedge for the existence of convex tax function, but hedging increases corporate debt capacity, which brings in 1.1% of firm value worth of tax benefits.

2.1.3 Insufficient investment

If cost of external financing is much more than cost of internal financing, hedging will be valuable because it can match cash inflows and outflows well and make it less likely to borrow from capital market. This will guarantee enough internal capital to be

(14)

invested in programs with positive present value and mitigate or eliminate capital insufficiency (Froot, Scharfstein, and Stein, 1993).

Gezy, Minton and Schrand (1997) finds the usage of interest rate derivatives is positively related company’s Research and Development expenses, based on a sample of Fortune Global 500. Gay and Nam (1998), Allayannis and Ofek (2001), Graham and Rogers (2002) also concludes that hedging increases the level of Research and Development expense.

Besides, it is very likely that companies with high grow potential and high debt ratio will hedge (Gezy, Minton, and Schrand, 1997).

2.1.4 Manger’s incentive

Generally, enterprise senior managers will make hedging strategies. From the perspective of managers, the optimal hedging strategy might depend on their compensation contracts.

Smith and Stulz (1985) claims in a two-period model if manager’s end-of-period wealth is a concave function of the end-of-period firm value, the optimal strategy is hedging; if it is a convex function, the optimal strategy is not hedging.

2.2 Motivations of state-owned firm speculating

2.2.1 Principal-agent problem

Chinese owned firms have a multi-level proxy mechanism. Theoretically, state-owned firms are funded and state-owned by all citizens. In other words, every one of the 1.3 billion taxpayers is a shareholder of state-owned enterprises. However, it is impossible to manage directly for everyone since taxpayers are too scattered. In 2003, after a system reform, state-owned companies established state-owned property sponsor system, specifying the responsibilities of sponsor agency. Now, there are more than

(15)

2000 listed companies in China. More than half of them are held by government and managed by State-owned Assets Supervision and Administration Commission of the State Council (SASAC) or local state-owned assets management departments, performing the agent responsibilities. It would be difficult to manage so many firms at the same time for SASAC, an impersonal entity (Han, and Zhang, 2009).

Then SASAC commissioned board of directors to manage firms and directors employ professional managers in charge of daily operation and management. Almost all directors and supervisors are managers from original state-owned firms, which means managers and board of directors are persons act in concert, except for very few independent or external directors. The corporate governance structure seems complete and sound, but fails to check and balance on each party. Considering that many directors also hold a post as manager, members of both board of directors and management are agents and insiders of state-owned enterprises. Insider control may occur because of information asymmetry (Li, and Liao, 2011).

2.2.2 Interest incentives of agents

A lot of state-owned firms have earned fat profits before huge losses occurred. Delivered return for Air China and China Eastern Airlines from fuel derivatives once exceeded main operating income. The exposure of fuel derivatives had increased from 30% of spot goods purchase amount to 50%. 2008 annual reports of these two airlines reveal that payment for directors, supervisors and senior managers is linked to company’s earnings.

The salary of executive director of CITIC Pacific is related positively to firm’s annual profits. The bonus accounts for approximately 2/3 of the annual payment. Yung Chi-kin, chairman of the board of directors, had bonus of HK$ 40 million and HK$ 48 million respectively in year 2006 and 2007. There was no bonus for him in year 2008 because of the big losses in derivatives (CITIC Pacific annual report 2006, 2007, and 2008).

(16)

2.2.3 No punishment for agents

A. Subsidies to loss-making enterprises

Brean (1998) writes in his book, “The most visible problem with Chinese enterprises is the substantial number running on loss … The problem of loss-making enterprises would seem to be due to three factors: first, a rise in the share of labor cost in value-added, second, increased competition in the product market, and third, rising debt service payments … There are two interrelated aspects to the subsidy for loss-making enterprise. First, enterprise face a soft-budget constraint, including the absence of the threat of bankruptcy, which has an important effect on enterprise behavior. They are likely to undertake more risky investments if they know that they will be subsidized when they make a loss.”

B. No responsibilities for losses

To avoid responsibilities of huge losses, agents usually insist that the derivative transactions are hedging or they have made poor judgment (Li, and Liao, 2011).

(17)

Section 3: Hypotheses

Flowing from the aims of the study, the hypotheses being tested and their underlying reasoning are summarized below.

Hypothesis 1: Sate-owned firms use less derivatives.

According to the literature review about motivations of employing derivatives and Subsidies to loss-making enterprises, government will inject capital if state-owned firms lose (Brean, 1998). It is less likely for state-owned enterprises to confront financial distress because of the support and protection from government. Therefore, state-owned firms may lack of the motive of reducing financial distressing cost and have less incentive to use derivatives.

Hypothesis 2: State-owned companies intend to use derivatives to speculate and so are

riskier than non-state-owned companies.

After analyzing the actual managers’ situation in both good and bad scenarios of using derivatives, we can infer that the agents of state-owned firm would harvest profits from gain in derivatives but lose nothing even derivatives perform terrible. From this point of view, state-owned firms have incentive to speculate.

(18)

Section 4: Data and descriptive statistics

4.1 Data collection

The study window is year 2014, in which we can review the latest annual reports for listed companies. And we will use yearly cross-sectional data.

We use Flush iFinD Financial Database to get most of data which I need for the study. The initial database consists of all Chinese manufacturing firms that were listed on Shanghai Stock Exchange and Shenzhen Stock Exchange during 2014. By using Flush iFinD, I can easily and fast access to financial statements and various financial analysis indicators, like total assets, liabilities to assets ratio, quick ratio, controlling shareholder of companies, variance of stock returns, beta in trading days over some specific period and so on.

The sample for the models consists of 340 listed manufacturing companies. In China, a large amount of derivative using companies are in manufacturing industry, including metal and nonmetal, oil, chemical engineering, plastic, machinery and instruments, food and drinks, pharmaceuticals, biomedicine, textile and clothing, wood and furniture and other industries (Zhao, 2011). Then I chose to use information of listed companies in manufacturing industry.

The extraction of data from Flush iFinD Financial Database resulted in an initial sample of 362 observations. Of these observations, 22 was removed because of missing data.

All information about whether the companies use derivatives is hand collected. The sources used to collect this information are off-balance entries and notes on annual

reports posted on website www.cninfo.com.cn, where China Securities Regulatory

Commission publish official data and announcements about securities market. In China, detailed information of financial derivatives assets and liabilities, like exposures of derivatives or what kinds of derivatives (interest, equity, currency, commodity or credit)

(19)

they are employing, is not obligated to be recorded in financial statements, which is not the same in the U.S.

4.2 Defining variables

4.2.1 Explained and explanatory variables

In Model 1 (Logit model, we will elaborate it later), the explained variable is whether

a firm uses derivatives. We can use dummy variables 𝑑𝑑𝑑𝑑 to represent the binary

variables derivative user or non-user. If the firm uses derivatives, 𝑑𝑑𝑑𝑑 = 1. And if the

firm does not use derivatives, 𝑑𝑑𝑑𝑑 = 0.

𝑑𝑑𝑑𝑑 = �10 if the firm is derivative userif the firm is non − user

In Model 1, the explanatory variable is whether the firm is state-owned. Similarly, we

can use dummy variables 𝑠𝑠𝑠𝑠𝑠𝑠 to represent the binary variables derivative user or

non-user. If the firm uses derivatives, 𝑠𝑠𝑠𝑠𝑠𝑠 = 1. And if the firm does not use derivatives,

𝑠𝑠𝑠𝑠𝑠𝑠 = 0.

𝑠𝑠𝑠𝑠𝑠𝑠 = �10 if the firm is state − ownedif the firm is non − state

We will judge whether a firm is state-owned by its controlling shareholder.

Here is the legal definition of controlling shareholder given byThe Company Law of

the People's Republic of China: a controlling shareholder generally controls the composition of the board of directors and influences the corporation’s activities (Company Law of the People's Republic of China, revised in 2005, Article 217). Sometimes, a shareholder who owns a smaller percentage but a significant number of remaining shares in the company can also be a controlling shareholder

(20)

In Model 2a (OLS regression), the explained variable is total risk measured by annualized standard deviation of stock returns. In Model 2b (OLS regression), the explained variable is systematic risk measured by beta, as in the capital asset pricing model (CAPM).

4.2.2 Control variables

Several proxies for factors which have an impact on firm risk are included as control variables.

A. Firm size

According to the capital structure theories, bigger firms with diversified operations have stronger anti-risk capacity. And smaller ones with uncertain operations usually cannot resist risk well. We take the natural logarithm of total asset as control variable 𝑠𝑠𝑠𝑠.

B. Leverage

According to the financial management theories, a larger leverage means a high ratio

of debt and carries more risks. We take liabilities-to-assets ratio 𝑠𝑠𝑎𝑎 as the proxy for

leverage.

C. Liquidity

Hedging theories state that firms with sufficient internal fund do not need to borrow from external capital market and can undertake projects of positive net present value (NPV) (Froot, Scharfstein, and Stein, 1993). Firms with higher liquidity will be less

risky. Quick ratio 𝑞𝑞𝑞𝑞 is the proxy for liquidiy.

D. Profitability

Generally, profitable firms is less risky. We adopt ROE 𝑞𝑞𝑟𝑟𝑑𝑑 as proxies for profitability,

(21)

considering that the high correlation of ROA and liabilities-to-assets ratio (Table 1) may invite multicollinearity.

Table 1. Correlation of ROA and liabilities-to-assets ratio

𝑠𝑠𝑎𝑎 𝑞𝑞𝑟𝑟𝑠𝑠

𝑠𝑠𝑎𝑎 1

𝑞𝑞𝑟𝑟𝑠𝑠 -0.624*** 1

*_{p < 0.05,}**_{p < 0.01,}***_{p < 0.001}

E. Growth opportunities

Companies with high grow potential will hedge (Gezy, Minton, and Schrand, 1997). Stock price is the sum of discounted future cash flows, and PE ratio is the price of a share divided by earnings per share. Therefore, PE ration can represent a firm’s future

growth opportunities. PE ratio 𝑝𝑝𝑑𝑑 will be proxy for growth opportunities.

4.3 Summary statistics

In this paragraph the summary statistics are displayed.

Table 2a summarizes the characteristics of derivative users and non-users. The summary statistics indicate that the two subgroups differ considerably on several dimensions.

The mean total assets for derivative users and non-users are ¥29.3 and ¥10.5 billion, respectively. And the median total assets for derivative users and non-users are ¥12.5 and ¥4.6 billion, respectively. The smaller size of the non-users and larger size of derivative users suggests that scale economies are associated with the adoption of a derivatives-based risk-management program. Besides, derivative users have higher beta, higher ROE and lower PE ratio than non-users.

Table 2b summarizes the characteristics of state-owned firms and non-state ones. The 21

(22)

summary statistics indicate that the two subgroups differ considerably on several dimensions.

State-owned firms have significantly bigger firm size, higher leverage, higher stock-return volatility, higher beta, and lower ROE than non-state firms after comparing means and medians of both subgroups.

Table 3 displays the correlation coefficients and their significance of each pair of variables. There is no correlation coefficient which not equals to zero significantly, except between liabilities-to-assets ratio and quick ratio. However, the absolute value of coefficient of -0.62 is less than 0.8, which means there is no need to worry about multicollinearity among variables (Ho, and Wong, 2001). All variables can be taken into analysis.

(23)

Table 2a. Summary statistics of derivative users and non-users

Derivative Users

Non-Users

Mean Median St. Dev. Min Max Mean Median St. Dev. Min Max

Total assets

2.93e+10 1.25e+10 4.58e+10 8.26e+08 2.29e+11 1.05e+10***

(-3.3933)

4.16e+09*** (-3.965)

2.69e+10 9.25e+07 4.15e+11

Liabilities-to-assets ratio 55.51988 58.10015 16.49585 10.7588 78.3189 50.22437 (-1.3371) 49.86365 (-1.527) 21.06912 3.5121 99.607 Quick ratio 1.080583 .9682 .9941343 .3499 6.065 1.277519 (0.7464) .8612 (0.016) 1.410982 .0685 11.8685 Volatility 35.01791 33.8954 5.865421 26.187 52.0433 36.87547 (1.2100) 35.7926 (1.270) 8.203945 19.9182 99.1564 Beta .8046733 .80935 .2071336 .3544 1.21 .7197506* (-1.8156) .71855** (-1.998) .2479143 -.0371 1.4699 23

(24)

ROE 10.04083 7.927 7.497868 .54 27.07 -4.537831 (-0.6208) 4.26*** (-2.969) 128.4392 -2228.37 102.97 PE ratio 143.5901 31.2861 473.9612 -35.6976 2593.061 197.1651 (0.4305) 61.9881** (2.399) 665.1267 -341.0863 8423.273 N 30 312

Notes: t-statistics in parentheses in Mean columns correspond to a student’s t-test of a difference in means between the Derivative Users and the sample denoted in the column heading. z-statistics in parentheses in Median columns correspond to a Wilcoxon test of a difference in medians between the Derivative Users and the sample denoted in the column heading. *denotes significance at the 0.10level. **denotes significance at the 0.05 level. ***denotes significance at the 0.01 level

Table 2b. Summary statistics of state-owned firms and non-state firms

State-owned

Non-state

Mean Median SD Min Max mean Median SD Min Max

Total assets

1.49e+10 4.62e+09 3.71e+10 1.06e+08 4.15e+11 8.33e+09**

(-2.0465)

3.94e+09*** (-2.823)

1.26e+10 9.25e+07 7.50e+10

Liabilities- 52.42085 52.3742 20.68803 7.3438 99.607 48.33582* 46.4941* 20.6548 3.5121 92.3297

(25)

to-assets ratio (-1.8058) (-1.767) Quick ratio 1.190189 .8681 1.236941 .0747 10.3941 1.355422 (1.0950) .8724 (0.506) 1.551697 .0685 11.8685 Volatility 37.24302 36.6102 7.205116 21.9479 58.7112 35.99177 (-1.4251) 34.6681** (-2.052) 9.020825 19.9182 99.1564 Beta .7861619 .7804 .2265439 .1318 1.4699 .6470931*** (-5.3849) .6555*** (-5.072) .2483507 -.0371 1.4585 ROE .3797624 3.7 24.25864 -191 77.44 -8.2027 (-0.6384) 5.84** (2.446) 186.6444 -2228.37 102.97 PE ratio 204.907 62.3256 588.0648 -333.0168 5702.556 175.5623 (-0.4119) 49.4721 (-1.327) 728.0339 -341.0863 8423.273 N 197 145

Notes: t-statistics in parentheses in Mean columns correspond to a student’s t-test of a difference in means between the Derivative Users and the sample denoted in the column heading. z-statistics in parentheses in Median columns correspond to a Wilcoxon test of a difference in medians between the Derivative Users and the sample denoted in the column heading. *denotes significance at the 0.10 level. **denotes significance at the 0.05 level. ***denotes significance at the 0.01 level

(26)

Table 3. Correlation matrix of variables 𝑠𝑠𝑠𝑠 𝑠𝑠𝑎𝑎 𝑞𝑞𝑞𝑞 𝑣𝑣𝑟𝑟𝑎𝑎 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 𝑞𝑞𝑟𝑟𝑑𝑑 𝑝𝑝𝑑𝑑 𝑠𝑠𝑠𝑠 1.000 𝑠𝑠𝑎𝑎 -0.052 1.000 (0.785) qr -0.122 -0.620*** 1.000 (0.520) (0.000) 𝑣𝑣𝑟𝑟𝑎𝑎 -0.285 0.013 0.250 1.000 (0.127) (0.947) (0.183) 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 0.534** -0.072 -0.019 0.079 1.000 (0.002) (0.704) (0.920) (0.680) 𝑞𝑞𝑟𝑟𝑑𝑑 -0.120 -0.312 0.245 -0.093 -0.159 1.000 (0.528) (0.093) (0.193) (0.623) (0.401) 𝑝𝑝𝑑𝑑 -0.057 0.216 -0.072 0.286 0.055 -0.236 1.000 (0.766) (0.251) (0.704) (0.125) (0.773) (0.210) p-values in parentheses *_{p < 0.05,}**_{p < 0.01,}***_{p < 0.001} 26

(27)

Section 5: Methodology

5.1 Model 1: Logistic regression

Logistic regression is commonly used to estimate the probability of a binary response based on one or more independent variables. Dependent variable is always assigned to

be 0 or 1. 𝑝𝑝_𝑖𝑖 represents the probability that the dependent variable equals a case, given

the values of independent variables. 𝑥𝑥_𝑖𝑖 is independent variable, and 𝑏𝑏_𝑖𝑖 is the

coefficient of 𝑥𝑥_𝑖𝑖.

The model is as follows,

𝑝𝑝𝑖𝑖 = 𝐸𝐸(𝑦𝑦 = 1|𝑥𝑥𝑖𝑖) = 1 1 + 𝑑𝑑−�𝑏𝑏0+∑𝑛𝑛𝑖𝑖=1𝑏𝑏𝑖𝑖𝑥𝑥𝑖𝑖� 𝐿𝐿 = ln_{1 −}𝑝𝑝𝑖𝑖_𝑝𝑝 𝑖𝑖=𝑏𝑏0+�𝑏𝑏𝑖𝑖 𝑛𝑛 𝑖𝑖=1 𝑥𝑥𝑖𝑖+ 𝜀𝜀

Logistic regression is an improvement of linear probability model (LPM). When 𝑥𝑥

change from negative infinity to positive infinity, probability 𝑝𝑝 change from 0 to 1.

However, the relationship between 𝑥𝑥 and 𝑝𝑝 is not linear anymore, which means we

cannot use OLS to estimate coefficients. Maximum likelihood estimation will be employed in logit model.

In our case, 𝑝𝑝_𝑖𝑖 = 1 means the firm uses derivatives, and 𝑝𝑝_𝑖𝑖 = 0 means the firm is

non-user. Our model is as follows,

ln_{1 −}𝑝𝑝𝑖𝑖_𝑝𝑝

𝑖𝑖=𝑏𝑏0+ 𝑏𝑏1∙ 𝑠𝑠𝑠𝑠𝑠𝑠𝑖𝑖+ 𝑏𝑏2∙ 𝑠𝑠𝑠𝑠𝑖𝑖+ 𝑏𝑏3∙ 𝑠𝑠𝑎𝑎𝑖𝑖+ 𝑏𝑏4∙ 𝑞𝑞𝑞𝑞𝑖𝑖+ 𝑏𝑏5∙ 𝑞𝑞𝑟𝑟𝑑𝑑𝑖𝑖+ 𝑏𝑏6∙ 𝑝𝑝𝑑𝑑𝑖𝑖+ 𝜀𝜀

If the coefficient 𝑏𝑏₁ is significant and positive, state-owned firm has higher probability

of engaging in derivative transactions than non-state firm. State-owned firms are tended to use more derivatives

(28)

If the coefficient 𝑏𝑏₁ is significant and positive, state-owned firm has lower probability of using derivatives than non-state firm. State-owned firms are tended to use more derivatives

If the coefficient 𝑏𝑏₁ is not significant, it shows no clear tendency of state-owned firms’

behavior.

5.2 Model 2: OLS regression

To test the interactive effect of using derivatives and state-owned firms, interaction term

of dummy variables 𝑑𝑑𝑑𝑑 and 𝑠𝑠𝑠𝑠𝑠𝑠 is introduced to the OLS regression.

5.2.1 Model 2a

In the first regression, we take total risk 𝑣𝑣𝑟𝑟𝑎𝑎 as explained variable.

Regression 1:

𝑣𝑣𝑟𝑟𝑎𝑎𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽1∙ 𝑑𝑑𝑑𝑑𝑖𝑖 + 𝛽𝛽2∙ 𝑠𝑠𝑠𝑠𝑠𝑠𝑖𝑖 + 𝛽𝛽3∙ 𝑑𝑑𝑑𝑑𝑖𝑖 ∙ 𝑠𝑠𝑠𝑠𝑠𝑠𝑖𝑖+ 𝛽𝛽4∙ 𝑠𝑠𝑠𝑠𝑖𝑖+ 𝛽𝛽5∙ 𝑠𝑠𝑎𝑎𝑖𝑖 + 𝛽𝛽6∙ 𝑞𝑞𝑞𝑞𝑖𝑖+ 𝛽𝛽7

∙ 𝑞𝑞𝑟𝑟𝑑𝑑𝑖𝑖 + 𝛽𝛽8∙ 𝑝𝑝𝑑𝑑𝑖𝑖 + 𝜇𝜇𝑖𝑖

Ignoring other factors, if the firm is state-owned (𝑠𝑠𝑠𝑠𝑠𝑠_𝑖𝑖 = 1), total risk 𝑣𝑣𝑟𝑟𝑎𝑎_𝑖𝑖 = 𝛽𝛽₁∙ 𝑑𝑑𝑑𝑑𝑖𝑖+ 𝛽𝛽2+ 𝛽𝛽3∙ 𝑑𝑑𝑑𝑑𝑖𝑖 = 𝛽𝛽2+ (𝛽𝛽1+ 𝛽𝛽3) ∙ 𝑑𝑑𝑑𝑑𝑖𝑖 and the effect of using derivatives is 𝛽𝛽1+

𝛽𝛽3,; if the firm is state-owned (𝑠𝑠𝑠𝑠𝑠𝑠𝑖𝑖 = 0), total risk 𝑣𝑣𝑟𝑟𝑎𝑎𝑖𝑖 = 𝛽𝛽1∙ 𝑑𝑑𝑑𝑑𝑖𝑖 and the effect of

using derivative is 𝛽𝛽₁. The coefficient of interaction 𝛽𝛽₃ estimates the effect difference of state-owned and non-state firms using derivatives, and shows whether using derivatives will contribute more risks after switching from subgroup Non-state to subgroup State-owned.

𝛽𝛽3 = 𝜕𝜕

2_{𝑣𝑣𝑟𝑟𝑎𝑎}

𝜕𝜕𝑠𝑠𝑠𝑠𝑠𝑠 𝜕𝜕𝑑𝑑𝑑𝑑

If 𝛽𝛽₃ is significant and positive, state-owned firms use derivatives to speculate. If 𝛽𝛽₃

(29)

is significant and negative, state-owned firms use derivatives to hedge.

5.2.2 Model 2b

Regression 2:

𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠𝑖𝑖 = 𝛼𝛼 + 𝛽𝛽1∙ 𝑑𝑑𝑑𝑑𝑖𝑖 + 𝛽𝛽2∙ 𝑠𝑠𝑠𝑠𝑠𝑠𝑖𝑖 + 𝛽𝛽3∙ 𝑑𝑑𝑑𝑑𝑖𝑖∙ 𝑠𝑠𝑠𝑠𝑠𝑠𝑖𝑖 + 𝛽𝛽4 ∙ 𝑠𝑠𝑠𝑠𝑖𝑖 + 𝛽𝛽5∙ 𝑠𝑠𝑎𝑎𝑖𝑖 + 𝛽𝛽6∙ 𝑞𝑞𝑞𝑞𝑖𝑖 + 𝛽𝛽7

∙ 𝑞𝑞𝑟𝑟𝑑𝑑𝑖𝑖 + 𝛽𝛽8∙ 𝑝𝑝𝑑𝑑𝑖𝑖 + 𝜇𝜇𝑖𝑖

Schrand, Unal (1998) found that corporations can control non-systematic risk better than systematic risk because of comparative advantage, and corporations will take risks to gain additional returns in operations with comparative advantage and use derivatives to hedge risks in operations without comparative advantage and reduce systematic risk.

Therefore, we replace total risk 𝑣𝑣𝑟𝑟𝑎𝑎 by systematic risk 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 in Model 2b and still test the interaction coefficient 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠.

If 𝛽𝛽₃ is significant and positive, state-owned firms use derivatives to speculate. If 𝛽𝛽₃

is significant and negative, state-owned firms use derivatives to hedge.

There is a likely simultaneity problem here. Risky firms may use derivatives to manage risks and more risks may invite higher probability of using derivatives. In Section 7, we would discuss it in detail and use instrumental variable to solve the endogeneity problem.

(30)

Section 6: Empirical results

6.1 Model 1: Logistic regression

The main logistic regression results of Model 1 are reported in Table 4.

The coefficient of 𝑠𝑠𝑠𝑠𝑠𝑠 is not significant. We cannot conclude from this result that

state-owned firms are tended to use less derivatives.

The only variable which has significant coefficient is firm size 𝑠𝑠𝑠𝑠. The coefficient is

positive and significant at the level of 1%, which means bigger firms are more likely to use derivatives. It is consistent with economic of scale hypothesis of firm size.

Table 4. Model 1 regression results

𝑑𝑑𝑑𝑑 Exponentiated coefficient 𝑠𝑠𝑠𝑠𝑠𝑠 -0.564 (-1.36) 𝑠𝑠𝑠𝑠 0.635*** (3.82) 𝑠𝑠𝑎𝑎 0.007 (0.49) 𝑞𝑞𝑞𝑞 0.015 (0.06) 𝑞𝑞𝑟𝑟𝑑𝑑 0.019 30

(31)

(1.60) 𝑝𝑝𝑑𝑑 -0.000 (-0.23) 𝑐𝑐𝑟𝑟𝑛𝑛𝑠𝑠𝑠𝑠𝑠𝑠𝑛𝑛𝑠𝑠 -16.980*** (-4.49) N 340 pseudo R2 0.115 t statistics in parentheses *_{p < 0.1,}**_{p < 0.05,}***_{p < 0.01}

6.2 Model 2: OLS regression

6.2.1 Model 2a

The coefficient of interaction 𝛽𝛽₃ is not significant and we cannot infer whether

state-owned firms use derivatives to increase or decrease risks.

The variables with significant coefficient are dummy variable 𝑠𝑠𝑠𝑠𝑠𝑠, firm size 𝑠𝑠𝑠𝑠, quick

ratio 𝑠𝑠𝑎𝑎, ROE, and PE ratio 𝑝𝑝𝑑𝑑.

The coefficient of 𝑠𝑠𝑠𝑠𝑠𝑠 is positive, which means state-owned firms have more total

risks. The coefficient of 𝑠𝑠𝑠𝑠 is negative, which means bigger firms face less total risks.

It is consistent with the assumption of firm size that bigger firms have greater anti-risk

capacity. Positive coefficient of 𝑠𝑠𝑎𝑎 shows that higher leverage carries more risks to the

firm. ROE has negative effect on total risk, which seems reasonable that firms with stronger profitability is less risky. Positive coefficient of PE ratio suggest that stock returns are more volatile if the company have greater growth opportunities.

6.2.2 Model 2b

The coefficient of interaction 𝛽𝛽₃ is negative and significant at the level of 10%, which

(32)

means state-owned firms use derivatives to reduce risks or hedge against systematic risks. However, we may arrive to an arbitrary conclusion if we terminate our research here.

The coefficients of 𝑑𝑑𝑑𝑑 and 𝑠𝑠𝑠𝑠𝑠𝑠 are positive and significant at the level of 10% and 5%

respectively. For non-state-owned firms, the marginal effect of using derivatives on systematic risk is 0.1071. We have presumed derivatives can successfully reduce risks. However, just as mentioned above, another possibility is that higher risk would

stimulate firms to use derivatives to hedge risks and the interpretation of 𝛽𝛽₃ may

ignore the simultaneity. Then we will perform additional test of endogeneity in Section 7.

Besides, the coefficients of firm size 𝑠𝑠𝑠𝑠, profitability 𝑞𝑞𝑟𝑟𝑑𝑑, and growth opportunities

𝑝𝑝𝑑𝑑 are significant at the level of 1%, 5% and 5% respectively. Positive coefficient of 𝑠𝑠𝑠𝑠 suggests that the larger the firm, the more systematic risk it would take. ROE and

PE ratio has negative and positive effects respectively on systematic risk 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠, which

seems reasonable that firms with stronger profitability and greater growth opportunities is less risky.

𝑣𝑣𝑟𝑟𝑎𝑎 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 Coefficient Coefficient 𝑑𝑑𝑑𝑑 1.258 0.107* (0.60) (1.78) 𝑠𝑠𝑠𝑠𝑠𝑠 1.772** _0.126*** (1.99) (4.95) 𝑑𝑑𝑑𝑑 ∙ 𝑠𝑠𝑠𝑠𝑠𝑠 -3.260 -0.154* (-1.12) (-1.85) 32

(33)

𝑠𝑠𝑠𝑠 -1.485*** 0.0725*** (-4.19) (7.14) 𝑠𝑠𝑎𝑎 0.0785*** -0.00123 (2.86) (-1.57) 𝑞𝑞𝑞𝑞 0.0846 -0.00543 (0.21) (-0.47) 𝑞𝑞𝑟𝑟𝑑𝑑 -0.00732** _-0.000231** (-2.15) (-2.37) 𝑝𝑝𝑑𝑑 0.00248*** 0.0000397** (3.80) (2.13) 𝑐𝑐𝑟𝑟𝑛𝑛𝑠𝑠𝑠𝑠𝑠𝑠𝑛𝑛𝑠𝑠 64.34*** -0.904*** (8.37) (-4.11) N 340 340 adj. R2 _0.113 _0.214 t statistics in parentheses *_{p < 0.1,}**_{p < 0.05,}***_{p < 0.01} 33

(34)

Section 7: Additional tests

7.1 Model 3: Logistic regression

To test whether risky firms are motivated to hedge risks with derivatives, the probability

of variable 𝑑𝑑𝑑𝑑 = 0 is adopted as explained variable. 𝑝𝑝_𝑖𝑖 = 1 means the firm uses

derivatives, and 𝑝𝑝_𝑖𝑖 = 0 means the firm is non-user. Our model is as follows,

𝑖𝑖=𝑏𝑏0+ 𝑏𝑏1∙ 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠𝑖𝑖+𝑏𝑏2∙ 𝑠𝑠𝑠𝑠𝑠𝑠𝑖𝑖+ 𝑏𝑏3∙ 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠𝑖𝑖∙ 𝑠𝑠𝑠𝑠𝑠𝑠𝑖𝑖+ 𝑏𝑏4∙ 𝑠𝑠𝑠𝑠𝑖𝑖+ 𝑏𝑏5∙ 𝑠𝑠𝑎𝑎𝑖𝑖+ 𝑏𝑏6∙ 𝑞𝑞𝑞𝑞𝑖𝑖

+ 𝑏𝑏7∙ 𝑞𝑞𝑟𝑟𝑑𝑑𝑖𝑖+ 𝑏𝑏8∙ 𝑝𝑝𝑑𝑑_𝑖𝑖+ 𝜀𝜀

The regression results are shown in Table 6.

The coefficients of systematic risk 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 and dummy variable 𝑠𝑠𝑠𝑠𝑠𝑠 are not significant,

which means the main effects of systematic risk 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 and dummy variable 𝑠𝑠𝑠𝑠𝑠𝑠 are

not significant. But the interaction effect is significant at the level of 10%. The effect

of systematic risk 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 on the probability of using derivatives is dependent on the

value of 𝑠𝑠𝑠𝑠𝑠𝑠.

𝑑𝑑𝑑𝑑 Coefficient 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 2.013 (1.59) 𝑠𝑠𝑠𝑠𝑠𝑠 1.719 (1.25) 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 ∙ 𝑠𝑠𝑠𝑠𝑠𝑠 -2.983* (-1.76) 34

(35)

𝑠𝑠𝑠𝑠 0.591*** (3.07) 𝑠𝑠𝑎𝑎 0.007 (0.46) 𝑞𝑞𝑞𝑞 0.018 (0.07) 𝑞𝑞𝑟𝑟𝑑𝑑 0.023* (1.70) 𝑝𝑝𝑑𝑑 -0.000 (-0.39) _cons -17.443*** (-4.20) N 340 pseudo R2 0.134 t statistics in parentheses *_{p < 0.1,}**_{p < 0.05,}***_{p < 0.01}

7.2 Model 4: Logistic regression

Now we split the whole sample into two subgroups 𝑠𝑠𝑠𝑠𝑠𝑠 = 0 and 𝑠𝑠𝑠𝑠𝑠𝑠 = 1 and do

logistic regressions respectively.

𝑖𝑖=𝑏𝑏0+ 𝑏𝑏1∙ 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠𝑖𝑖+ 𝑏𝑏2∙ 𝑠𝑠𝑠𝑠𝑖𝑖+ 𝑏𝑏3∙ 𝑠𝑠𝑎𝑎𝑖𝑖+ 𝑏𝑏4∙ 𝑞𝑞𝑞𝑞𝑖𝑖+ 𝑏𝑏5∙ 𝑞𝑞𝑟𝑟𝑑𝑑𝑖𝑖+ 𝑏𝑏6∙ 𝑝𝑝𝑑𝑑𝑖𝑖+ 𝜀𝜀

Table 7 illustrate the regression results of Model 4. 35

(36)

For state-owned firms, the effect of systematic risk 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 on the probability of using derivatives is negative and not significant. There is no need to test simultaneity for subgroup State-owned.

For non-state firms, the effect of systematic risk 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 on the probability of using

derivatives is positive and significant at the level of 10%. Simultaneity problem may occur in this subgroup.

𝑑𝑑𝑑𝑑 (𝑠𝑠𝑠𝑠𝑠𝑠 = 0) 𝑑𝑑𝑑𝑑 (𝑠𝑠𝑠𝑠𝑠𝑠 = 1) Coefficient Coefficient 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 2.342* -1.382 (1.68) (-0.98) 𝑠𝑠𝑠𝑠 0.566* 0.657** (1.94) (2.49) 𝑠𝑠𝑎𝑎 -0.004 0.002 (-0.17) (0.10) 𝑞𝑞𝑞𝑞 0.018 -0.370 (0.07) (-0.76) 𝑞𝑞𝑟𝑟𝑑𝑑 0.026 0.024 (1.20) (1.28) 𝑝𝑝𝑑𝑑 -0.004 0.001 (-1.09) (1.36) 36

(37)

_cons -16.417*** -16.463***

(-2.67) (-2.96)

N 143 197

pseudo R2 0.193 0.118

t statistics in parentheses *_{p < 0.1,}**_{p < 0.05,}***_{p < 0.01}

7.3 Instrumental variable (IV)

Instrumental variable method are commonly applied to solve endogeneity problem. A suitable instrumental variable should meet two requirements:

A. The instrument must be correlated with the endogenous explanatory variables.

B. The instrument should be from outside of the system, that is, instrument should have exogeneity.

The second requirement is especially necessary for simultaneity problem. Although the correlation between instrumental and endogenous variables can be estimated in statistics, the causal chains behind causal relationship of instrumental and endogenous variables must depend on convincing logical derivation (Acemoglu, and Johnson et al., 2001; Angrist, 2008). The exogeneity of instrumental variables means IV is uncorrelated with the error term in the explanatory equation. In other words, IV will not relate to explained variable directly. Otherwise, it is only endogenous variable that can connect IV and explained variable.

In our study, a suitable IV should predict systematic risk 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 well and affect

company’s derivative using decision only through 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠. From this point of view,

institutional ownership level is a good IV. On one hand, institutional ownership is correlated to beta. Badrinath et al. (1989) identify a performance of institutional investors for high beta. Holding a stock with a high beta increases the expected return

(38)

of the stock, thereby decreasing the probability of sub-standard portfolio performance, which suggest a positive relationship between beta and institutional ownership. As

Table 8 demonstrate, the correlation coefficient of 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 and institutional ownership

𝑖𝑖𝑟𝑟 is significant at the level of 1%.

Table 8. Correlation between beta and institutional ownership

𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 𝑖𝑖𝑟𝑟

𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 1.000

𝑖𝑖𝑟𝑟 0.154*** 1.000

(0.004)

p-values in parentheses *_{p < 0.1,}**_{p < 0.05,}***_{p < 0.01}

On the other hand, using derivatives is decided inside a company and would not be affected by external institutional investors. In addition, contract of derivatives is not revealed to the public, and there is only very limited information announced in annual reports. External institutional investors are unlikely to base their portfolio strategies on

whether a company use derivatives. The variable institutional ownership 𝑖𝑖𝑟𝑟 can be

viewed as exogenous and from outside the system of an enterprise. Two-step probit regression is established to perform the Wald test for exogeneity. The explained variable is 𝑃𝑃𝑞𝑞𝑟𝑟𝑏𝑏. (𝑑𝑑𝑑𝑑 = 1) and the explanatory variable is 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠.

From Table 9, there are no variables with significant coefficients. According to the Wald test for exogeneity, the statistic is insignificant, which means: there is no endogeneity for non-state firm’s logistic regression in Model 4

It implies that for state-owned firms, riskier companies are more likely to use derivatives.

(39)

Table 9. Two-step probit with endogenous regressors Coefficient Std. Err. 𝑧𝑧 𝑃𝑃 > 𝑧𝑧 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 -30.56398 92.5936 -0.33 0.741 𝑠𝑠𝑠𝑠 2.834074 7.385521 0.38 0.701 𝑠𝑠𝑎𝑎 -0.0830988 0.2384946 -0.35 0.728 𝑞𝑞𝑞𝑞 -0.2571951 0.9122787 -0.28 0.778 𝑞𝑞𝑟𝑟𝑑𝑑 0.0066896 0.0269694 0.25 0.804 𝑝𝑝𝑑𝑑 -0.0012088 0.0030447 -0.4 0.691 𝑐𝑐𝑟𝑟𝑛𝑛𝑠𝑠𝑠𝑠𝑠𝑠𝑛𝑛𝑠𝑠 -40.08428 91.54345 -0.44 0.661 Instrumented: 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 Instruments: 𝑠𝑠𝑠𝑠 𝑠𝑠𝑎𝑎 𝑞𝑞𝑞𝑞 𝑞𝑞𝑟𝑟𝑑𝑑 𝑝𝑝𝑑𝑑 Wald test of exogeneity:

chi2(1) = 1.83 Prob. > chi2 = 0.1767

(40)

Section 8: Conclusion

Just as Hull (2006) presents, two different behaviors of corporation using derivatives are hedging and speculating. As studies of motivations of hedging show that hedging is motivated by the needs of decreasing risk and then increasing firm value, which is a long-term financial objective. If firms only focus on short-run operating performance, they will not be risk-averse anymore and try to undertake risks in order to harvest short-term returns. Li and Liao (2011) analyze principal-agent problem in Chinese state-owned firms and the cost and return of agents speculating, which suggest that agents can harvest substantial profit from earnings of derivatives and cost nothing if the company lose in derivative market. Brean (1998) analyzes subsidies from government to lose-making companies in China and concludes the absence of the threat of bankruptcy has an important effect on state-owned enterprise behavior. They are likely to undertake more risky investments if they know that they will be subsidized when they make a loss.

This paper examines whether state-owned enterprise use derivatives to speculate rather than to hedge risk as they claim.

Cao’s (2013) conclusion of ineffective risk management in Chinese corporations implying speculating rather than hedging is inconsistent with conclusions of Guay (1999), Wong (2000) and Bartram, Brown, and Conrad (2009), whose studies are based on more mature U.S. market.

Therefore, it is necessary to test in China whether derivative users are hedging or whether their hedging strategy is useful before we test whether state-owned companies are riskier than non-state-owned ones.

Two hypotheses are constructed. And we need to first test whether state-owned firms have less incentive to use derivatives and then test whether state-owned enterprise are tended to use derivatives speculatively.

(41)

Logistic model is developed to test the relationship of probability of derivative-use and whether the company is state-owned. The coefficient of explanatory dummy variable 𝑠𝑠𝑠𝑠𝑠𝑠 (whether the derivatives using company is owned by government) is not significant, so Hypothesis 1 cannot be proved.

Model 2a and Model 2b explain annualized standard deviation stock returns (total risk)

and CAPM beta (systematic risk) respectively, using dummy variables 𝑑𝑑𝑑𝑑 (whether the

company use derivatives during study window), 𝑠𝑠𝑠𝑠𝑠𝑠 and their multiplicative

interaction term as explanatory variables and controlling other firm characteristics. The

coefficient of interaction term 𝛽𝛽₃ is not significant in Model 2a and significantly

negative in Model 2b. It means state-owned firms have stronger reducing effect of derivative-use on systematic risk and they are hedging against systematic risk (rather than total risk) by using derivatives.

Besides, we notice that non-state firms’ marginal effect of derivative-use on systematic risk is 0.1071, which enable us to question our presumption that derivative-use can reduce risks. The reverse effect may also exist that more risks spur to company’s derivative-use decision. The simultaneity problem have to be tested.

Then we swap explained variable 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 for explanatory variable 𝑑𝑑𝑑𝑑 and construct

another logistic regression model to test the marginal effect of systematic risk 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 on

the probability of using derivatives, which is found to rely on the value of 𝑠𝑠𝑠𝑠𝑠𝑠.

State-owned firms’ effect is negative and not significant, which means the effect of derivatives reducing systematic risk dominate. Meanwhile, non-state firms have

significantly positive effect of 𝑏𝑏𝑑𝑑𝑠𝑠𝑠𝑠 on the probability of 𝑑𝑑𝑑𝑑 = 1. It suggest that this

effect may dominate and endogeneity should be tested for non-state firms.

Instrumental variable institutional ownership 𝑖𝑖𝑟𝑟 is correlated to systematic risk beta

logically and statistically. In addition, institutional ownership will not directly affect a firm’s decision to use derivatives for risk management and vice versa. After a two-step probit regression on beta and control variables, the Wald test of exogeneity shows

(42)

insignificance, which means no endogeneity in logistic regression ln_{1 −}𝑝𝑝𝑖𝑖_𝑝𝑝

for non-state firms. The effect of risk inviting derivative-use decision dominate the direction of the relationship between systematic risk beta and derivative-use for non-state firms.

The overall conclusion is that Chinese state-owned firms are tended to use derivatives to hedge against systematic risks.

(43)

Reference

Acemoglu, D., Johnson, S., & Robinson, J. A. (2001). The colonial origins of comparative. The Economics of Cultural Transmission and the Evolution of Preferences,” Journal of.

Allayannis, G., & Ofek, E. (2001). Exchange rate exposure, hedging, and the use of foreign currency derivatives. Journal of international money and finance, 20(2), 273-296.

Angrist, J. D., & Pischke, J. S. (2008). Mostly harmless econometrics: An empiricist's companion. Princeton university press.

Badrinath, S. G., Gay, G. D., & Kale, J. R. (1989). Patterns of institutional investment, prudence, and the managerial" safety-net" hypothesis. Journal of Risk and Insurance, 605-629.

Bartram, S. M., Brown, G. W., & Conrad, J. (2011). The effects of derivatives on firm risk and value. Journal of Financial and Quantitative Analysis, 46(04), 967-999.

Benston, G. J., & Mian, S. L. (1995). Financial reporting of derivatives: An analysis of the issues, evaluation of proposals, and a suggested solution. J. OF FINANCIAL ENGINEERING, 4(3).

Brean, D. J. (2013). Taxation in modern China. Routledge

Cao, Y. (2013). The risk management on derivatives application: Evidence from China’s listed Company. The Theory and Practice of Finance and Economics, 34(5).

Carter, D., Rogers, D. A., & Simkins, B. J. (2004). Does fuel hedging make economic sense? The case of the US airline industry. The Case of the US Airline Industry (September 16, 2002). AFA.

(44)

Fok, R. C., Carroll, C., & Chiou, M. C. (1997). Determinants of corporate hedging and derivatives: A revisit. Journal of Economics and Business,49(6), 569-585.

Froot, K. A., Scharfstein, D. S., & Stein, J. C. (1993). Risk management: Coordinating corporate investment and financing policies. the Journal of Finance, 48(5), 1629-1658.

Fu, P. (2011). Chinese derivatives and accounting problems. Financial Theories and Practice, (04), 11-13.

Gay, G. D., & Nam, J. (1998). The underinvestment problem and corporate derivatives use. Financial Management, 53-69.

Géczy, C., Minton, B. A., & Schrand, C. (1997). Why firms use currency derivatives. The Journal of Finance, 52(4), 1323-1354.

Gogineni, S. (2010). Oil and the stock market: An industry level analysis.Financial Review, 45(4), 995-1010.

Graham, J. R., & Rogers, D. A. (2002). Do firms hedge in response to tax incentives?. The Journal of Finance, 57(2), 815-839.

Guay, W. R. (1999). The impact of derivatives on firm risk: An empirical examination of new derivative users. Journal of Accounting and Economics,26(1), 319-351.

Haushalter, G. D. (2000). Financing policy, basis risk, and corporate hedging: Evidence from oil and gas producers. Journal of Finance, 107-152.

Hentschel, L., & Kothari, S. P. (2001). Are corporations reducing or taking risks with derivatives?. Journal of Financial and Quantitative Analysis,36(01), 93-118.

Hull, J. C. (2006). Options, futures, and other derivatives. Pearson Education India.

Jimenez-Rodriguez, R. (2008). The impact of oil price shocks: evidence from the 44

(45)

industries of six OECD countries. Energy Economics, 30(6), 3095-3108.

Jin, Y., & Jorion, P. (2006). Firm value and hedging: Evidence from US oil and gas producers. The Journal of Finance, 61(2), 893-919.

Jorion, P. (1990). The exchange-rate exposure of US multinationals. Journal of business, 331-345.

Kim, Y., Nam, J., & Wynne, K. (2009). An event study approach to shocks in gold prices on hedged and non-hedged gold companies. Investment Management and Financial Innovations, 6(2), 112-119.

Koski, J. L., & Pontiff, J. (1999). How Are Derivatives Used? Evidence from the Mutual Fund Industry (Digest Summary). Journal of Finance, 54(2), 791-816.

Lee, K., & Ni, S. (2002). On the dynamic effects of oil price shocks: a study using industry level data. Journal of Monetary Economics, 49(4), 823-852.

Leland, H. E. (1998). Agency Costs, Risk Management, and Capital Structure (Digest Summary). Journal of finance, 53(4), 1213-43.

Li, H., & Liao, X. (2011). Research on failure of state-owned firms using financial derivatives to hedge risks. International Business Accounting, (4), 55-59

Lin, R., & Chang, Y. (2009). Does hedging add value? Evidence from the global airline industry. Evidence from the Global Airline Industry (February 13, 2009).

Li, Y., & Xiao, H. (2010). Derivatives and company risks: An empirical study. Financial Theories and Practice, (09), 39-43.

Modigliani, F., & Miller, M. H. (1958). The cost of capital, corporation finance and the theory of investment. The American economic review, 261-297.

(46)

Nance, D. R., Smith Jr, C. W., & Smithson, C. W. (1993). On the determinants of corporate hedging. Journal of finance, 267-284.

Scholtens, B., & Yurtsever, C. (2012). Oil price shocks and European industries. Energy Economics, 34(4), 1187-1195.

Seiler, M. J. (2004). Performing financial studies: a methodological cookbook. Prentice Hall.

Smith, C. W., & Stulz, R. M. (1985). The determinants of firms' hedging policies. Journal of financial and quantitative analysis, 20(04), 391-405.

Stulz, R. M. (1984). Optimal hedging policies. Journal of Financial and Quantitative Analysis, 19(02), 127-140.

Stulz, R. M. (1996). Rethinking risk management. Journal of applied corporate finance, 9(3), 8-25.

Zhao, X. (2011). Derivative usage and firm value, Risk: Evidence from Chinese nonferrous metals listed companies. Economic Management, Vol.33, No.1.

An empirical analysis on whether Chinese state-owned firms are tended to use derivatives speculatively