The Chinese Accounting Standards (CAS) in 2006 included Hedging Accounting Standards for the first time, which made it possible to do a large sample of Chinese listed companies related to the use of derivative. However, in 2007, the subprime mortgage crisis rapidly spread into a global financial crisis, and the use of derivative became the hardest hit, and the use of derivative by Chinese firms was also quite affected. Derivative use gradually returned to normal after the post-crisis era in 2009. Based on this, this paper takes A-share listed companies from 2010-2016 as the initial sample to study the impact of derivative use on companies’ investment level.

Firms with no risk exposure during the sample period are deleted from this paper as the use of derivative is mainly used to hedge against the risk of exchange rate, interest rate and commodity price fluctuations. Exposures are judged as follows: a company is considered to have foreign exchange risk exposure if it has foreign income or foreign subsidiaries (Makar & Huffman, 2001) ; Commodity exposure is typically industry-specific, and an industry is considered to have commodity exposure if the derivatives users percentage of commodity derivative exceeds 10 per cent (Purnanandam, 2008) ; It is also considered that firms that use interest rate derivatives and other derivatives are also exposed. Since financial sector firms are both users and providers of derivatives, this paper removes the financial sector firms; And due to the higher probability of financial anomalies in ST companies, companies that were ST during the sample period and those with missing data were removed, resulting in 2083 companies and 10,710 observations.

Drawing on Triki (2006) , Chang et al. (2015) , and Guo et al. (2021) , data on the use of derivative are collected manually in firms’ annual reports in the following way: Search for derivatives, hedges, forwards, futures, options, swaps, swaps, financial assets (liabilities) held for trading, assets (liabilities) at fair value through profit or loss, other (non) current assets (liabilities), etc., gains and losses on changes in fair value, investment income, and risks associated with (related to) financial instruments in the annual report, and determine whether the company uses derivatives, the fair value of derivatives and the impact of t derivatives on the gains and losses on changes in fair value and on investment income, etc., according to the context.

Foreign currency statement translation differences data from Reset database, R&D related data from CCER database, and other financial data from Wind database.

To test the effect of derivative usage on the level of investment, following Froot et al. (1993) , Berrospide et al. (2011) , model (1) is constructed:

Capx i t + 1 = α 0 + α 1 DT i t + Controls i t + ∑ Industry + ∑ Year + ε i t (1)

Capx_it+1 is the capital expenditure level, and following Trapp and Weiβ (2016) , capital expenditure is measured by the proportion of cash paid for the purchase and construction of fixed assets, intangible assets and other long-term assets in operating income. In order to mitigate the possible reverse causality between derivative use and investment, the effect of derivative use in period t on the level of investment in period t + 1 is examined. DT is a dummy variable for the use of derivative and takes the value of 1 if derivatives are used and 0 otherwise.

Controls are control variables including firm size (Size), tangible assets (PPE), profitability (ROA), gearing ratio (Debtratio), quick ratio (Quickratio), firm’s financial crisis likelihood (Z-value), non-debt tax shield (SGA), firm’s growth (MB), share of overseas business (Oversea), Dibble’s internal control index (Icindex), nature of ownership (Ownership), firm’s age (Age), cash dividend per share (Dividend), and net cash flow from operating activities as a percentage of operating income (OCF).

Industry and Year are industry and year control variables respectively, to control the impact of different investment opportunity in different industries, macro-economy such as GDP and inflation on capital expenditure. The definition and measurement of variables are shown in Table 1.

The results of the descriptive statistics are shown in Table 2, and as can be seen from Panel A, the average proportion of capital expenditure to total assets is 11.6%. The mean of Tobin Q 2.372, the average proportion of R&D expenditure to operating revenue is 10%.¹ During the sample period, 21.1% of companies used derivatives. The proportion of derivative users is much lower than the proportion of developed countries such as Europe and the United States, as well as developing countries such as South Africa and Malaysia. For derivative users, the fair value of derivative accounts for 10% of total assets. In terms of control variables, on average, the proportion of fixed assets to total assets is 21.5%, the return on total assets is 5.64%, the mean gearing ratio is 40.8%, the mean quick ratio is 2.267, the mean Z-value is 8.417, the sum of administrative expenses and selling expenses as a percentage of operating income is 16.2%, the mean MB of the company is 2.802, the mean percentage of overseas business is 17.5%, the mean value of the Dibble Internal Control Index is 607.6, 33.1% of companies are SOEs, the mean cash dividend per share is 0.119, and the mean value of the company’s net cash flow from operating activities as a percentage of operating income is 7.59%.

As can be seen from Panel B, the number of derivative users increased year by year, from 216 in 2010 to 431 in 2016. The proportion of derivative users also showed an overall upward trend, from 12.4% to 16.2%. And as can be seen from Panel C, the mean and median values of capital expenditure are lower for derivative users compared to firms that do not use derivatives. The T-value for the test of difference in means is 7.8429 and the Z-value for the median rank sum test is 8.221, respectively, which are both significant at the 1% level. The results of Panel C indicate that the capital expenditure level of derivative users is lower than non-users, supporting hypothesis H1b to some extent.

Before regression, this paper conducts correlation coefficient analysis on the main variables. The Pearson correlation coefficient matrix shows that the use of derivative is negatively related to the level of capital expenditure and significantly at the 1% level, which supports H1b to some extent too. Most of the correlation coefficients between the main independent variables are less than 0.3, with a maximum of 0.6, and the mean variance inflation factor (VIF) is 2.35, indicating that there is no serious covariance between the main independent variables.

To control for possible fixed effects and sequence related issues in companies, this paper uses the cluster regression method recommended by Petersen (2009) and Gow et al. (2010) to perform dual clustering by company and year.

To test hypothesis 1, model (1) is regressed and the results are shown in column (1) of Table 3, the derivative use dummy variable DT is significantly negatively correlated with firms’ capital expenditures at the 1% level, suggesting that the level of capital expenditures of derivative users is lower as compared to firms that do not use derivatives, and Hypothesis 1b passes the test.

a. The numbers in parentheses represent the T-value of the robust standard error for double clustering at the company and year levels; *, **, and *** represent significant at the 10%, 5%, and 1% levels, respectively.

Different from studies on companies in developed Western countries, the use of derivative not only do not improve the investment level of Chinese listed companies, but on the contrary, the use of derivative reduces the company’s capital expenditure level. From the perspective of control variables, consistent with previous studies, the level of capital expenditure is higher, when the total asset (Size) is larger, the proportion of tangible assets is higher, the company’s liquidity is better, the non-debt tax shield is higher, the company’s growth is better, the proportion of overseas business income is higher, and the internal cash flow is more abundant.

1) Replacement of dependent variable

In order to test the robustness of the research results, the dependent variable is replaced. Two measurement methods are adopted for the level of capital expenditure：The first approach is to use the sum of cash paid for the purchase and construction of fixed assets, intangible assets and other long-term assets and net cash paid for the acquisition of subsidiaries and other business units, less the net cash recovered from the disposal of fixed assets, intangible assets, investment properties and other long-term assets and the net cash received from the disposal of subsidiaries or other business units, the result is Capx1, which is the level of the company’s capital expenditure, and standardised by operating income; The second way to measure capital expenditure is by the level of R&D expenditure. R&D expenditure is also an important component of corporate capital expenditure, which is the sum of expensed and capitalized R&D expenditure, to operating income.

The regression results, as shown in columns (2) and (3) of Table 3, indicate that the dummy variable for derivative use DT, is still significantly negatively correlated with capital expenditures after replacing the capital expenditures measure, and both are significant at the 1% level, suggesting that the results of this paper are unaffected by the choice of the dependent variable indicators.

2) Endogeneity test: propensity score matching method

Bartram et al. (2011) use propensity score matching to control for the endogeneity problem of derivative use, and this paper draws on that approach to pair a sample of derivative users with a sample of non-users.

Based on Chang et al. (2015) , Donohoe (2015) , and Campbell et al. (2023) , the paired variables are factors affecting the use of derivative as well as factors affecting the level of investment, including market value of the firm (MV), gearing (Debt ratio), ROA of total assets, ROA volatility (ROAVOL), cash flow volatility (OCFVOL), firm bankruptcy likelihood (Z), firm growth (MB), cash dividends per share (Dividend), and net cash flow from operating activities (OCF), industry and year effects. In this article, the propensity score matching method uses logit to estimate propensity scores, and only matches individuals within a common value range. The specific pairing method is nearest neighbor one-on-one and put back matching.

The difference in capital expenditure of derivative users and non-users before and after pairing is shown in Table 4, from which it can be seen that after pairing, the level of capital expenditure of derivative users is lower, with a T-value of −5.89, which is greater in absolute value than the critical value of 1.96, indicating that the level of capital expenditure of derivative users is low, which, to some extent, indicates that the findings of this paper are robust.

Continuing the regression after pairing, the results are shown in column (4) of Table 3: consistent with the results of the main regression in this paper, after pairing, the use of derivative is significantly negatively correlated with the company’s capital expenditure level, and the findings are robust.

3) Endogeneity tests: treatment effects model

Derivative use is an endogenous choice for companies (Pincus & Rajgopal, 2002; Choi et al., 2015) , company size, asset-liability ratio, exchange rate, interest rate and commodity price exposure, potential for financial crisis, performance sensitivity to management compensation, tax, growth, profit and cash flow volatility all affect the use of derivative. For this reason, this paper employs a treatment effects model to mitigate the problem of self-selection of the sample. The process is described below: Examining the effect of factors influencing the use of derivative by constructing model (2), from which we get Inverse Mills Ratio λ_i, and then add to model (1) for correcting the effect of derivative use on the level of firms’ capital expenditures in the second-stage regression. If the coefficient of λ_i is significant, then there is indeed sample selection bias and the use of the treatment effects model is valid.

Pr ( DT i t = 1 ) = Φ ( α i Z i ) (2)

Z_i is the influencing factor for the use of derivative. Following on Chang et al. (2015) and Campbell et al. (2023) , firm size (Size), gearing ratio (Debt ratio), exchange rate exposure (F risk), interest rate exposure (I risk), commodity price exposure (C risk), financial crisis likelihood (Z), management compensation performance sensitivity (MI), cash effective tax rate (CETR), growth (MB), the volatility of return on total assets (ROAVOL), and cash flow volatility (NCFVOL) are included, and are lagged one period for all variables except firm size (Size).

The results are shown in Table 5: column (1) presents the results of the first-stage regression, in which firms that are larger and have higher debt ratios are more likely to use derivative. And column (2) is the regression results of derivative usage on the level of capital expenditure after the inclusion of λ_i, where λ_i is significant at the 1 percent level, suggesting that there is a self-selection problem and that the use of a treatment effects model is necessary; After accounting for the self-selection bias, the regression coefficients of the dummy variable for derivative use DT, remain significant at the 1% level, indicating that the level of capital expenditures by derivative users is significantly lower than that of firms that do not use derivative, and that the conclusion is robust.

a. The numbers in brackets are z-values; *, ** and *** represent significant at the level of 10%, 5%, and 1% respectively.

1) The impact of the risk management effect of derivatives.

The risk management effect of derivatives is reflected in the impact on the volatility of corporate cash flow. The fluctuation of macroeconomic factors such as exchange rate, interest rate and commodity price, will cause the fluctuation of internal cash flow of enterprises, and then lead to the fluctuation of external financing need or investment level. When cash flow is short, if there are investment projects with net present value greater than 0, additional external financing should be carried out, or the investment projects should be postponed or even cancelled. At this time, it is a better choice to carry out unplanned financing. However, according to the pecking order theory, the cost of external financing is higher than that of internal financing, which may make the project with net present value greater than 0 unfeasible, thus reducing the investment level. When the risk management of derivatives is effective, the derivatives usage will hedge the impact of fluctuations of macroeconomic factors on cash flow, thus reducing the volatility of the company’s cash flow, and stable cash flow will help the company to carry out investment and financing activities in an orderly manner. If on the contrary, it will have an adverse impact on capital expenditure behavior.

In order to further test the risk management effect of derivatives, this paper examines the impact of derivatives on cash flow volatility. The results are shown in column (1) of Table 6. Derivative usage dummy variable DT is positively correlated with cash flow volatility, but it does not pass the significance test, indicating that the use of derivative instruments does not play a role in reducing cash flow volatility.

2) The impact of derivatives usage on the ability to borrow.

The insufficient financing of firms may lead to a decrease in capital expenditure. The use of derivative could raise the level of capital expenditure by increasing the ability of companies to borrow and provide more funding for investment (Carter & Simkins, 2006) . If derivatives adversely affect the ability to borrow, it will reduce the level of capital expenditure.

To further examine the impact of derivative usage on debt capacity, following on Zou (2010) and Trapp and Weiβ (2016) , we adopt the ratio of company i’s new debt in year t + 1 to the total assets at the end of year t (Debtcap_it+1) to measure of the company’s debt capacity in year t. The results, as shown in column (2) of Table 6, show that derivatives and the level of new debt are significantly negative at the 1% level, indicating that derivative usage reduced the company’s debt capacity and further reduced the company’s capital expenditure level.

1) The impact of the derivative use intensity of on investment level

Following on Wong (2000) and Campbell et al. (2019) , the fair value of derivative to total assets (DT ratio) is used to measure the intensity of derivative use. On the basis of model (1), DT ratio is used to replace dummy variable DT for regression. The results are shown in column (1) of Table 7, and it is found that the intensity of derivative use (DT ratio) is negatively correlated with the level of capital expenditure, and is significant at the 1% level, indicating that among derivative users, the higher the proportion of fair value of derivatives to total assets, the lower the level of capital expenditure.

a. Figures in brackets are T-values of robust standard errors of double clustering at the company and year levels; *, ** and *** represent significant levels at 10%, 5% and 1% respectively.

a. Figures in brackets are T-values of robust standard errors of double clustering at the company and year levels; *, ** and *** represent significant levels at 10%, 5% and 1% respectively.

2) The impact of ownership

This paper continues to examine the impact of derivative usage on capital expenditure in different ownership companies. The results are shown in column (2) of Table 7, where the coefficient of is negative and significant at the 1% level, indicating that the use of derivative has reduced the capital expenditure level of non-SOE enterprises. The coefficient of the intersection term DT * Ownership is significantly positive at the 10% level, indicating that compared to SOEs, the derivative usage has a stronger effect on reducing capital expenditure in non-SOEs.

3) Impact of Fair Value Measurement and Presentation of Financial Instruments (Revised) in 2014.

In 2014, the Ministry of Finance revised the accounting standards for the Presentation of Financial Instruments and formulated the standards for Fair Value Measurement. The Presentation of Financial Instruments (revised in 2014) elucidates the objectives of presenting financial instruments: to facilitate a reasonable assessment of the significance of financial instruments to enterprises’ financial position and operating results, as well as the nature and extent of risks that these instruments expose enterprises to during the reporting period and at the end of the period. The “Fair value measurement” standard requires the disclosure of the three levels of fair value measurement and measurement basis, the standard emphasizes that the basis of confirmation is “contract terms and the economic substance reflected”, and “not only in legal form” to determine whether it is a derivative instrument, and in many places adds the use of cases that should be recognized as derivative instruments. This paper further examines the impact that changes to accounting standards related to derivatives in 2014 on the capital expenditure behavior of derivative users. Therefore, on the basis of model (1), model (3) is constructed:

Capx i t + 1 = α 0 + α 1 DT i t + α 2 DT i t ∗ POST 2014 + POST 2014                             + Controls i t + ∑ Industry + ∑ Year + ε i t (3)

The treatment group is the derivative users (DT = 1), control group for firms that do not use derivatives (DT = 0). The value of POST2014 from 2010 to 2013 is 0; From 2014 to 2016, the value of POST2014 is 1. Controls_it is the control variables, which are consistent with model (1).

The results are shown in column (3) of Table 7, where the coefficient of the Intersection term DT * POST2014 is positive and significant at the 5% level, indicating that after 2014, the level of capital expenditure by derivative users is relatively high compared to companies that do not use derivatives. In other words, the revision of accounting standards related to derivatives in 2014 increases the level of capital expenditure for derivative users, that is the accounting standard revision helps to alleviate the adverse impact of derivative on capital expenditure.

4) Derivative use and firm value.

As can be seen from the above results, the use of derivatives reduces the level of capital expenditure of the company, and may therefore reduce the value of the company. In order to test the impact of the use of derivatives on corporate value, following on Froot et al. (1993) and Berrospide et al. (2011) , we build model (4):

TO i t = α 0 + α 1 DT i t + Controls i t + ∑ Industry + ∑ Year + ε i t (4)

TQ_it is the Tobin Q value of firm i in year t.

As shown in column (1) of Table 8, the use of derivative is negatively correlated with the Tobin Q value of the company, that is, the company value of the derivative user is lower than that of the company that does not use derivative. However, as shown in column (2), the coefficient of the intersection term DT * POST2014 is positive and significant at the 10% level, which suggests that after the implementation of the Fair Value Measurement Standards and the presentation of Financial Instrument Standards in 2014, the corporate value of derivative users is higher than that of companies that do not use derivatives, indicating that the reform of accounting standards help to mitigate the adverse impact of derivative on corporate value.

Figures in brackets are T-values of robust standard errors of double clustering at the company and year levels; *, ** and *** represent significant levels at 10%, 5% and 1% respectively.