CSR and brand value

The current research holds two opposite views about the relationship between CSR and brand value. Some scholars consider that performing social responsibility is beneficial to promote the brand value of enterprises. First, Chinese enterprises still treat the performance of social responsibility as a type of “advertisement” marketing behavior [12], considering that association produced by CSR can deliver altruistic value of enterprises in a warmhearted manner, that is, satisfying customer and social demands, which is beneficial in obtaining brand reputation [13] and then enhancing brand value [14, 15]. Second, brand fulfilling social responsibility is favored by consumers, which can establish a strong emotional connection with consumers [16] and bring high market share to enterprises [17], which can positively influence brand value. Furthermore, the source of brand value is its future profitability, and it is the willingness and motivation of consumers to continuously pay for brand beyond product or service premium [18]. According to Demetriou et al., corporate measures supporting and facilitating social prosperity may increase additional value of current products/services [19]. Lerro et al. indicate that consumers usually tend to purchase enterprise products with good public praise of social responsibility [20]. CSR is not only helpful for enhancing the purchasing intention of consumers and improving public evaluation on enterprises/brands [21] but also beneficial for brand extension [22]. The research results of Lai et al. also show that CSR activities have a positive influence on brand performance [23]. In conclusion, CSR can win recognition from the public and stakeholders, enhance popularity and brand reputation [24, 25], and promote brand competitiveness on the market, thereby producing positive influence on brand value [26].

Other scholars think that the input of CSR hinders the promotion of brand value. At first, in the process of achieving improved performance, enterprises have extensive contact with stakeholders, which makes them pay attention to CSR. At the same time, they expect pressure in various ways, which makes enterprises bear the social responsibility of paying high costs, which crowd out the resources needed for brand value building [27, 28] and lead to the decline of brand value. Furthermore, if social responsibility behaviors of some enterprises are considered by the public negatively, such as a commercial show or moral hypocrisy, then they cannot bring a positive reputation asset to enterprises and even corrode corporate reputation [29]. If the field of CSR is only related with its business, stakeholders will think that CSR aims to eliminate negative externalities of its main business activity or gain high profits, which will bring negative effect to the brand [30]. By analyzing the panel data of A-share listed companies, Li et al. find that CSR has a negative inhibition effect on brand value [31]. Therefore, blind performance of social responsibility can lead to irrational resource allocation, and false social responsibility will damage brand value [32].

Government subsidy and CSR

Research conclusions about the promotion or inhibition effect of government subsidy on CSR by domestic and foreign scholars differ. Some scholars consider that government subsidy is a nonmarket means of the government to overcome the externality of the environment and facilitate environmental improvement, which promotes CSR inputs [33]. First, government subsidies provide a capital source for enterprises, relieve financing constraints, and improve corporate operation state [34]; thus, enterprises can undertake social responsibility. The research by Zeng et al. shows that government subsidies can expand the cash flow of enterprises, increase external financing scale, reduce financing costs, and effectively solve the problem of financing constraints during production and operation process [35]. After obtaining government subsidies, enterprises can deliver good development prospect to others and attract attention of external investment organizations, such as banks, thereby increasing the probability of obtaining short- and long-term bank loans and investment organizations [36, 37]. With a rich cash flow, enterprises can finish production and operation activities effectively, and they can and are willing to undertake social responsibilities. Second, obtaining political resources or lowering political costs is a powerful motivation for enterprises to fulfill social responsibility; that is, CSR activities have become a way to consolidate government–enterprise association [38]. With government subsidies, enterprises will initiatively assume social responsibility in return to support government to realize political or social goals [39]. Donations are relatively a direct way for enterprises to perform social responsibilities. Private enterprises utilize donations to consolidate government–enterprise association [40], and money-losing enterprises with political resources tend to obtain government support through charitable donations [41]. If state-owned enterprises with tight government–enterprise association do not assume social responsibility actively, then they will undertake higher political costs and may lose government support, such as preferential policy and resource subsidy [42]. To develop and maintain government–enterprise relationship and acquire continuous support from the government, enterprises are motived to repay the government and society through social responsibilities. Therefore, enterprises with government subsidies will fulfill social responsibility actively. On the one hand, government subsidies improve enterprise operation state, and enterprises, in turn, can and are willing to perform social responsibilities. On the other hand, government–enterprise association is enhanced, enterprises continuously obtain government subsidy resources, and increasing political costs caused by failure to perform social responsibilities is avoided.

Other scholars claim that government subsidies hinder CSR inputs. First, abuse of government subsidy funds reduces CSR inputs. Under the situation of weak constraints, government subsidies may induce moral risks. For example, some enterprises grant executive compensation with government subsidies, making it difficult to realize the goal of government subsidies [43–45]. Possible corruption and rent-seeking behavior during the allocation process of government subsidy funds hinder efficient allocation and weaken subsidy effect [46]. Moreover, government subsidies may influence enterprise operation performance, and implementing social responsibility activities is not beneficial. Money-losing listed companies can improve their financial condition in the year with government subsidies; however, enterprises will depend on government subsidy in the long term, which is not beneficial to improve operation performance [47]. Through government subsidies, controlling shareholders increase surplus while reducing earnings quality [48], which leads to inefficiency of government subsidy [49]. With poor earnings quality, few extra resources will be inputted for social responsibility activities. Consequently, government subsidies have a negative influence on CSR. On the one hand, abuse of government subsidy reduces CSR inputs. On the other hand, government subsidies lead to dependence of enterprise operation and no idle resources are inputted for CSR.

In conclusion, regarding the relationship between CSR and brand value and between government subsidy and CSR, most studies at home and abroad tend to focus on the perspective of the linear effect between variables, and no consistent conclusion exists at present. Actually, diversity of research conclusions indicates a complex relationship between CSR and brand value and between government subsidy and CSR to some extent, and complex nonlinear relationship may exist between them. What type of complex mechanism will government subsidy produce on the effect between CSR and brand value? What type of characteristics and rules will be produced? If a threshold exists, then how will the characteristics of the threshold be manifested? These issues need to be discussed further.

Model design

As per the panel threshold effect model proposed by Hansen [50], three threshold models with a single threshold value are initially constructed, and a multiple-threshold model is then gradually expanded. The three models of a single threshold value are as follows.

Model (1) is the threshold regression model, with CSR being the threshold variable and independent variable at the same time, with brand value being the dependent variable.

(1)

Model (2) the is regression model, with government subsidy being the threshold variable and independent variable at the same time, with CSR being the dependent variable.

(2)

Model (3) is the threshold model, with government subsidy being the threshold variable, CSR being the independent variable, and brand value being the dependent variable.

(3)

In the models, i refers to a specific company; t is the year of a company; corresponding coefficient vectors are expressed with β₁, β₂, and θ′; Sub_it, CSR_it, and BV_it refer to government subsidy, input of CSR, and brand value, respectively; and γ is a specific threshold value. X_it is a control variable, including enterprise size, financial leverage, shareholder balance, and return on total assets, which may obviously influence brand value and CSR. I(·) is an exponential function, μ_i represents the individual effect of the enterprise, and ε_it is a random disturbance term. Specific meanings and calculation methods of the variables are as follows.

Models (1)–(3) are combined as (4)

If Tv in Formula (4) is BV, then it is Models (1) and (3). If Tv is CSR, then it is Model (2). If Tr is CSR, then it is Models (1) and (3). If Tr is Sub, then it is Model (2). If Ts is CSR, then it is Model (1). If Ts is Sub, then it is Models (2) and (3).

To estimate the parameters, the interclass mean is subtracted from each observed value to eliminate the individual effect, and the model after conversion is (5)

All observed values are superimposed, and Formula (5) is expressed in matrix form, as shown as follows: (6)

For a given threshold value γ, the least square estimation of Formula (6) is adopted, and the value of β is obtained as follows: (7)

The corresponding residual sum of squares is (8)

Where is the residual vector. By using the minimum of Formula (8), the estimated value of γ can be obtained as (9)

Then, whether the threshold effect and the estimated value are equal to the true value is tested. The null hypothesis and corresponding alternative hypothesis of the threshold effect test are respectively as follows: H₀: β₁ = β₂ and H₁: β₁ ≠ β₂. The test statistics can be expressed as follows: (10)

Under the condition of the null hypothesis, the threshold value cannot be identified. As a result, F₁ statistics is an abnormal distribution. The null hypothesis of the second test is H₀: γ = γ₀, and the corresponding likelihood ratio statistics is (11)

Given that the distribution of the statistics is abnormal, this study refers to the formula proposed by Hansen (1999) to obtain the nonrejection region, that is, if LR₁ (γ₀) ≤ c(α), then the null hypothesis cannot be refused, in which c(α) = −2ln(1 − (1 − α)^1/2), where α indicates a significant level.

A threshold exists in the above hypothesis, namely, parameter estimation and hypothesis testing process of the single-threshold panel model. The dual model is expanded into a multiple model, as shown as follows: (12)

First, the estimated value of known is hypothesized and γ₂ is searched, which yields the following:

This study shows that is gradual and effective, whereas does not show this nature. Here, can be fixed, and is searched again to obtain a consistent statistic after optimization. Parameter estimation and hypothesis testing process of the multiple-threshold model are similar to those of the single-threshold model, as shown as follows: (13)

Sample and variable selection

The selection interval of the samples in this study is the listed companies in China’s Top 500 Most Valuable Brands issued by the World Brand Lab in 2012–2018, and the samples are screened as follows. (1) Enterprises among A-share listed companies in Shanghai and Shenzhen ranking for seven years are selected. (2) Financial and insurance companies are removed, because their statement structures and main titles of account are different from those of the general industry, and their business models also differ from other types of enterprise. (3) Samples under abnormal transaction status are removed, including ST, SB, *ST, and PT share. (4) Missing value sample is removed. After screening, the cross-section data of 110 sample companies are finally obtained, with a total of 770 samples.

The explained variable is brand value (BV), and the natural logarithm of brand value data in China’s Top 500 Most Valuable Brands issued by the World Brand Lab is used for measurement. Government subsidy (Sub) and the input strength of CSR (CSR) are threshold variables, and CSR is also an independent variable. With regard to government subsidy, the natural logarithm of government subsidy amount received by enterprises is used for measurement. For the measurement of the input strength of CSR, this study refers to the method of Shen et al. to measure with social contribution per share [51]. Social contribution per share = (net profit + business tax and surcharges + donation fee + financial expense + income tax expense–pollution discharge and clearing expense + cash paid to staff and paid for staff + employee pay payable in current period − employee pay payable in prior period)/mean of total shares at the beginning and end of the period. The control variables are the following: enterprise size (size, natural logarithm of enterprise total assets); financial leverage (Lev, gross liabilities/total assets); return on total assets (Roa, net profit/total assets); shareholder balance (Shb, sum of ratio of the second to tenth largest shareholders/shareholding amount of the largest shareholder); shareholding ratio of board of directors (Bsr, total shareholding of board members/total share of company); ratio of independent directors (Indep, number of independent directors/total number of board members); and duality (Dual, Dual = 1, duality; Dual = 0, not duality).