CSR is a self-regulatory business model that enables a firm to be socially responsible to itself, shareholders, and the general public (Wang & Zhou, 2020). Companies become aware of their influence on all elements of society, including the economic, social, and environmental facets by adopting CSR (Chia, Kern & Neville, 2020). Zhang et al. (2021) argue that firm engagement in CSR allows companies to introduce and promote value and help maintain a good reputation in the market (Johnson, Lee & Ashoori, 2018). Yoon, Chastagner and Joo (2020) conclude that CSR helps firms have a strong connection with the customer, which increases firm market shares and customer willingness to pay, irrespective of the services and quality of the product (Du et al., 2007). Vuković, Miletić, Čurčić and Ničić (2020) believe that a firm's engagement in social responsibilities helps promote the product, as it positively affects consumers’ sentiments, and the firm/product can enjoy a good reputation in the market (Lerro, Raimondo, Stanco, Nazzaro & Marotta, 2019). Torelli, Monga and Kaikati (2012) remark that firms with CSR have more opportunities to extend their markets than those who do not engage in such activities. By contrast, Christensen, Morsing and Thyssen (2020), Zhang et al. (2018), and Qiang et al. (2019) suggest that performing CSR activities builds business reputation, but also puts extra pressure on firms to perform well to maintain position, which increases product costs.

Despite Shepherd's (1972) declaration that political considerations with all types of subsidies are essential in determining firms' market positions, the question of whether subsidies improve enterprises' market power and affect competition remains debatable. Several factors may lead to a positive correlation between subsidies and firms' market power. First, as per Khurshid (2015), the number of subsidies reduces the firm's costs and uplifts the working capital, which provides a competitive edge and leads to unfair competition in the market. Similarly, Feldman and Kelley (2006) state that subsidies help firms secure loans and funding from financial institutions and banks and leads to other unseen benefits. Second, according to Aghion and Howitt (1992) and Klette and Griliches (2000), the beneficial effect of subsidies on innovation may result in a positive link between subsidies and market power given that innovation has been shown to be a key source of monopolistic power (Liu & Huang, 2016). In contrast, government-subsidised businesses have higher costs and less market influence. First, as (Zhao et al., 2015) note, firms in China seeking subsidies build and manage their relationship with officials, which increases firm costs and wastes resources that they can use to improve business performance. Yu, Hui and Pan (2010) demonstrate that financial subsidies to China's local government-owned firms will negatively impact its efficiency. Liang, Li and Lv (2012) note that subsidies in developing countries usually attract inefficient enterprises to participate in the marketplace. In addition, managers commit financial fraud and manipulate numbers linked to the labour force and productivity for more funds. Second, subsidies weaken the motivation for supported firms to increase efficiency since they may readily enjoy consistent and often considerable advantages. According to Lin (2012), from a viability perspective, large subsidy payments are made to businesses when the government wants these industries to grow, violating the factor endowment advantage. Businesses that lack viability will find it difficult to compete internationally if they lose the subsidy. Therefore, their market power and competition are linked to a continuous supply of government subsidies. Such cash flows in a market with a high degree of competition can diminish the responsiveness of businesses to the pressure of competition, which can lead to the so-called ‘producing just for subsidies’ practice and oversupply. Therefore, a company's market power and competition relations rely on the proportion of these two features when a subsidy increases or decreases.

Few studies focus on the relationship between CSR and government subsidies. According to one perspective, subsidies represent a tool to control environmental externalities which pave the way for CSR (Lu, Ren, Lin, He & Streimikis, 2019).

However, Li, Wu and Jiang (2018) and Khurshid et al. (2021) remark that subsidisation policies are a source of capital for private firms and help them finance their expenditures. As Guo, He and Xiao (2011) note, this also creates an opportunity for private firms to access banks and investment organisations for loans and investments. With the availability of cash, firms can contribute effectively by increasing their efficiency and initiating social responsibility. Second, this act of engagement in social activities strengthens the enterprise-government relationship. Jia and Liu (2014) argue that firms willingly engage in CSR after receiving subsidies. An interesting perspective is put forward by Li et al., who note that the government provides subsidies to achieve its political goals (Li et al., 2018). Donation is another way to strengthen the association between government and private firms (Xue & Xiao, 2011). In this regard, Lee, Walker and Zeng (2017) note that if firms do not engage in social responsibility, they bear high political costs, especially in the case of government enterprises. Therefore, private enterprises actively participate and perform social responsibility to repay the government.

Some studies argue that government subsidisation policies hinder CSR contributions. According to M. Shao and Bao (2012), the structural and institutional weakness on the government side puts a big question mark on allocating and utilising subsidy funds. These funds impact enterprises' operational performance, which further hinders their social responsibility. Moreover, Tzelepis and Skuras (2004) claim that subsidies are helpful for firms that run with a deficit, but this affects their performance because of extra pressure in the long run and decreases their profit. Firms' productivity can be enhanced by incorporating additional inputs; however, this increases the costs. Similarly, the orthodox mindset of firms motivates them to not perform social duties, as it will reduce revenue (Khurshid et al., 2020). Jia, Nam and Chun (2021) conclude that subsidisation policies increase firms’ dependence, as they do not use idle resources to maintain corporate social responsibly. This discussion shows that there is no clear consensus among scholars, which indicates the complexity of the relationship between the two concepts.

The research questions investigated in this study are novel relative to the recent literature. To the best of our knowledge, no study has considered market competition, the characteristics of the life cycle of firms, the number of subsidies, their volume, industrial preferences, and political connections. Unlike previous research, this study uses the FE2SLS method, which combines instrumental variables (IV) and fixed effects (FE). Therefore, both methods produce reliable results (Lin & Wooldridge, 2019: 24). It addresses multicollinearity, heteroskedasticity, unobserved exogenous effects, and selection bias problems and minimises the ambiguity of the outcome. This method is rarely used in empirical literature, especially in the CSR context. The ordinary least squares (OLS) method with robust standard error settings is applied for the robustness of the outcomes, as it helps minimise the error better than normal settings. The development of the model, selection of critical variables, and analysis of the relationships in different contexts differentiate this study different from the literature.

The theoretical model developed in this study is based on an oligopolistic, competitive market. We assume that two firms with similar products have imperfect substitutes. Both firms compete for market share and positions. Suppose Qi and Qj are the outputs produced by firms i and j, respectively. Thus, the total output in the market is Q=Qi+Qj. There are diminishing returns to scale as far as cost is concerned; therefore, the cost functions for both firms are C(Qi)=Qi2and C(Qj)=Qj2. Suppose that the market size is standardised to 1, and in such a situation, the market demand function is

The relationship between the government and private firms has been established in the literature (Wang, Khurshid, Qayyum & Calin, 2022), which significantly impacts firms’ decision-making processes (Ahmed & Javed, 2017; Joshi & Rahman, 2015). The literature also supports the negative consequences of firm employees’ political connections on productivity (Abbas & Awan, 2017), which help firms gain government subsidies. Based on this discussion, the current study follows Tomaru and Saito (2010), who introduce an exogenous parameter Subi (0 < Subi < 1) to indicate the degree of government subsidies to enterprises. The higher the Subi, the larger the subsidy provided by the government (M. Shao & Bao, 2012). Thus, the profit earnings of firms i,j can be represented by Eq. (1) as follows:

Furthermore, as Aggarwal and Samwick (2006) specify, each company has abundant resources to make independent decisions (SID) to maximise its profits, as shown below:

Any firm's primary objective is to maximise profits. The best-response function is estimated by partial differentiation of (1) for firms i and j and equating the system to zero (∂δi(Qi,Qj)∂Qi=∂δj(Qi,Qj)∂Qj=0) following the guidelines of Eaton and Grossman (1986) to obtain the equilibrium quantity (Qi and Qj). That is:

To obtain the equilibrium profit of the firms, we substitute Eq. (3) into Eq. (1), and after simplifying, we obtain Eq. (4):

To obtain the CSR function, we merge Eqs. (4) and (2) and take the first derivative with respect to government subsidies to establish the relationship between the two variables, in line with the proposition introduced by Song, Yan and Yao (2020).

Eq. (5) shows the direct and positive relationship between government subsidisation policies and corporate social responsibilities. This also shows that firm investment decisions (SIDs) are not subject to changes in government subsidies. Therefore, we propose the following hypotheses:

H1: Keeping the other variables constant, government subsidisation policies help private enterprises accomplish their CSR goals.

H2: Keeping the other variables constant, government subsidies discourage market competition.

By comparing Eqs. (2) and (4), we express the product market competition. We establish the association between product market competition and government subsidisation policies through a first-order partial derivative, as shown in Eq. (6).

Eq. (6) shows that government subsidies discourage competition.

H3: Keeping the other variables constant, growing market competition negatively affects private firms’ CSR.

To prove this proposition, we take the first-order partial derivative of our equation of CSRi of a private firm to γ. This leads to an association between CSRi and product market competition, that is,

In the next stage, to establish the relationship between CSR and government subsidies, we follow Tomaru and Saito (2010) and combine Eqs. (5), (6), and (7), as represented in (8), in the following manner:

Eq. (8) shows that government subsidies use product market competition as a transmission channel to minimise competition and provide an environment in which firms can perform CSR effectively.

The Karachi Stock Exchange is the largest market in Pakistan, with 552 registered companies in 36 different sectors. The listed companies engage in various actions oriented toward health, education, environment, and national disaster relief under the slogan of a sustainable living plan. In 2019, 57 companies were honoured by international CSR awards.1 This study selects the top 100 companies due to data availability constraints. These companies are from the manufacturing, services, financial, hospitality, real estate, and information technology sectors. The heterogeneous influence of the industry attributes of enterprises is tested by dividing the sample into two sub-samples: competitive (manufacturing) and non-competitive industries (others). The sample ranges from 2011 2019. The data on these firms are taken from OpenDoor.pk (https://opendoors.pk/premium-data/corporate-social-responsibility-csr-data-of-pakistan-psx-listed-firms/) and the Corporate Social Responsibility Centre Pakistan (https://csrcp.com/). Data on government subsidies to private firms are taken from various issues of the Economic Survey of Pakistan (https://www.finance.gov.pk) and firms' financial reports.

The following econometric model is used to examine the impact of government subsidies on private firms’ CSR activities. This relationship is tested using the OLS method with a robust standard error setting. The ‘demean technique’ is used to estimate the fixed effects IV regression (FE2SLS), and the estimator is referred to as a ‘within estimator.’ It also presumes that the relationship between the independent parameters and an individual's effect is known.

CSR denotes corporate social responsibility, Sub is the subsidy, cont is the control variable, Y represents the year, and Ind denotes the industry. Here, uit represents the time-invariant unobservable effects that can influence our dependent variables and εit represents the error term.

By government subsidy, we imply any policy form, either in cash or tax reduction, announced by the government that helps firms reduce their economic burden and promote social welfare. In this paper, we obtain our specific data for the measure of government subsidy by consulting the detailed item ‘government subsidy’ of ‘non-operating income’ in the notes to the financial statements of listed companies, manually collecting the amount, and taking the natural logarithm of the amount.

The selection of the control variables is based on previous studies (Khurshid et al., 2020; Zhang et al., 2021). Most studies considered equity, internal control quality index, size of the board of directors, size of the supervisory board, and corporate retained earnings to total assets. Other variables include growth opportunity (ratio between the market value of the firm and its total assets), total cost (internal), total size of the firm (total assets), age of the firm (years in business), and ratio between assets and liabilities. We also focus on measuring and controlling year FE to avoid any ambiguity. A winsorisation procedure is adopted to reduce the error caused by extreme values at the 1% and 99% quantiles. The descriptions of the variables are summarised in Table 1.

The theoretical derivation indicates that government subsidies mainly reduce the intensity of product market competition and ultimately encourage enterprises to actively perform social responsibility. Therefore, while keeping this in mind, this study constructs the following equations to empirically test the effect of government subsidies on the social responsibility of private enterprises.

In models (10) and (11), the meaning and measurement methods of variables such as CSRit, subsidy (Subit), and control (contit) are consistent with those in Model (9). However, COMPTit represents the product market competition. The Lerner index is more in line with the market concept for measurement, and is a reverse competition index. Therefore, this study uses its reciprocal as a proxy variable for product market competition, as suggested by Iqbal, Tauni and Jebran (2017). The specific calculation method is COMPTit which is as under.

In Model (9), β1 is the total effect of government subsidies on CSR without considering product market competition. In Model (10), α1 is the impact of government subsidies on product market competition. In Model (11), δ1is the impact of product market competition on CSR, and δ2 is the direct effect of government subsidies on CSR. After controlling for the degree of product market competition, α1δ1 is the mediating effect of government subsidies that affect CSR through the mediation variable of product market competition.

Specifically, the test procedure for the mediation effect is as follows: First, test Model (9). If β1 is not significant, the mediation effect test is terminated. However, if β1 is significant, the test models (10) and (11) are in sequence. Second, if both α1 and δ1 are significant, then the mediation effect (α1δ1) is significant. Under this condition, if δ2is significant, then there is a partial mediation effect. In addition, if at least one of α1 and δ1 are not significant, a Sobel test is required. If the test result is significant, it indicates a mediating effect; otherwise, it indicates no mediating effect. The Sobel (1982) effect is estimated using Z=α^1δ^1/Sα1δ1. where, α^1,δ^1are the estimated values for α1 and δ1. Sα1δ1=α^12Sδ12+δ^12Sα12, where Sα1, and Sδ1are the robust standard errors of α^1 and δ^1 respectively.

Descriptive statistical analyses of the main variables used in this study are summarised in Table 2. The results show that the company's overall social responsibility performance is 22.31. The standard deviation is 19.72, the minimum value is −3.58, and the maximum value is 61.63, indicating considerable differences in the level of fulfilment of social responsibilities among the sample companies. The government subsidies received by the companies appear unevenly distributed, and the average value after taking the logarithm is 18. We also observe a significant difference between the maximum and minimum values. The figure shows the differences in the quality of internal controls and retained earnings among companies. However, there are large differences in the internal costs. The largest shareholder holds more than 35% of the shares and the proportion of independent directors is approximately one-third. The average age of the company's listing is approximately six years, and the average asset-liability ratio is 36.2%.

The receipt of government subsidies may not be random for companies. For instance, government subsidies and their volumes may be based on a company's characteristics or industry. We directly select a sample of companies that have received government subsidies for estimation, which may cause a sample selectivity bias. In this regard, we use the Heckman two-stage model to make corrections. First, we construct dummy variables based on the strength of government subsidies and use the probit method in the first stage to estimate the probability of private enterprises receiving high-intensity government subsidies, that is, the inverse Mills ratio (IMR). Later, we add this as a control variable to the first stage in the two-stage regression. Column (1) of Table 4 summarises the results. The results show that subsidies contribute positively to CSR. In addition, the IMR is highly significant, indicating that the basic conclusions of this study are still valid after controlling for the sample self-selection problem. These results are consistent with those obtained using the OLS regression method.

Enterprises’ active charitable donations are a representative way for enterprises to fulfil their social responsibilities. Therefore, for the robustness of the results, we replace CSR with the natural log values of ‘public welfare donation expenditures’ under the label of ‘Non-operating Expenses’ – financial statements of listed companies. Similarly, to examine its nexus with CSR, we replace subsidy with its intensity (government subsidy amount/total corporate assets × 100). Column (2) of Table 4 presents the empirical relationship between these two variables. The outcomes show that, even after adjusting the measurement methods of the key variables, the overall regression results did not substantially change.

From a static perspective, the results show that government subsidies promote CSR significantly. Does this conclusion hold for companies with different life cycles? In this regard, this study divides the entire sample into three life stages–growth, maturity, and decline–and tests the benchmark model in dynamic settings. First, the top 50% of companies with positive operating income growth rates are classified as a growth period, and the last 50% are incorporated as the mature period. In contrast, companies with negative operating income growth rates were categorised as in the decline period. The results are shown in columns (3) – (5) of Table 5. The outcomes are consistent in both cases. The results show that, regardless of whether the sample enterprises are in the growth, maturity, or decline periods, there is a significant positive correlation between government subsidies and CSR.

The critical value of Sobel's test statistic at the 5% significance level is approximately 0.97. The regression results in Table 3 show that β1 in Model (9) is significantly positive; therefore, only the estimated coefficients of Models (10) and (11) need to be tested in turn. Table 6 lists the relevant test results.

The results in Column (1) show that government subsidies are negatively correlated with product market competition. The regression coefficient is −0.13, significant at the 1% level, indicating that government subsidies significantly reduced the intensity of product market competition. These results support the theoretical framework of this study. The results further reveal a significantly negative correlation between product market competition and CSR. A coefficient value of −0.21 indicates that product market competition reduces CSR. The regression coefficient of government subsidies in Column (2) is 0.49, which is significant at the 5% level. In summary, subsidies weaken the impact on CSR by reducing competition. The empirical results are in line with the theoretical framework developed in this study.

Considering that the government has considerable corporate heterogeneity in the number of subsidies, industry preferences for subsidies, political connections, and additional subsidies, this may affect the relationship between government subsidies and the social responsibility of private enterprises. In this regard, this section focuses on further analysis of the benchmark Model (9) from the perspective of the heterogeneity of government subsidies.

Financial subsidies provided by the government are not random. Corporate subsidies are typically based on the characteristics of enterprises or industries. Therefore, the number of subsidies received by different types of enterprises can vary significantly. In this regard, this study uses detailed information on government subsidies disclosed in the ‘non-operating income’ item in the notes to the annual statements of listed companies. The dataset is manually collected, the number of government subsidies obtained by enterprises is sorted, and the full sample is divided into averages according to government subsidies. The companies are divided into three main groups. The lowest frequency is defined as low-frequency arrays, the middle as medium-frequency arrays, and the highest groups are defined as high-frequency arrays and regressed as per our benchmark model. The test results are listed in Table 6 (rows 1–3). The effect of government subsidies on the social responsibility of private enterprises is only established in the low and middle-frequency sample group. Essentially, higher government subsidies do not play a role in fulfilling social responsibilities.

Mao and Xu (2018) argue that there should be a moderate range of government subsidies in which companies can be more productive. This division is based on the average values of company size and subsidies received. The lowest rage is the low-quota group, while the other two are the medium and high quota groups. Model 9 is tested in a new setting, and the outcomes are summarised in rows 4–6 of Table 6. The results show that medium-sized subsidies have a more significant and positive impact on CSR. In summary, the optimal range of government subsidies can only promote the CSR of enterprises.

To improve the general welfare of society, governments generally tend to provide more subsidies to enterprises with strong public attributes. Therefore, it is necessary to examine the heterogeneous influence of enterprise-industry attributes. This study divides the sample into two sub-samples of competitive industries (manufacturing) and non-competitive industries (others), and re-regresses the benchmark model. The results are summarised in rows 7–8 of Table 6. The findings reveal that government subsidies significantly impact CSR in competitive industry groups. Competition boosts the CSR activities.2

The connection between enterprises and government makes it easier for businesses to obtain government subsidies. However, this political connection can also lead to distortions in resource allocation and severe rent-seeking behaviour (Khurshid, Kedong, Calin & Popovici, 2016). For example, Faccio, Masulis and McConnell (2006) find that using political connections as an informal channel by enterprises to obtain government subsidies significantly reduces government subsidy resource efficiency. Fan, Wong and Zhang (2007) point out that securing government subsidies substantially improves operating efficiency and social contribution for enterprises that are not politically connected. Therefore, this study divides the companies in the sample into two sub-groups: with3 and without political connections. The test results are shown in rows (9) – (10) of Table 6. Non-political enterprises positively contribute to CSR. Enterprises that support political activities or are run and owned by politicians are not inclined toward CSR. The results of the FE2SLS and OLS analyses are consistent in both the intensity and direction of the relationship.