The Impact of On-the-Job Consumption on the Sustainable Development of Enterprises

: Under certain circumstances, on-the-job consumption is conducive to improving the resilience of the supply chain and the sustainable development ability of enterprises. Using China’s A-share listed companies from 2008 to 2019 as sample data in conjunction with the deconstructive analysis of on-the-job consumption, we used the multiple linear regression model of econometrics to deeply analyze the impact of on-the-job consumption on the sustainable development of enterprises, test the regulatory effect of innovation efﬁciency, and explore the governance effect of internal control and anti-corruption. Research shows that reasonable and excessive on-the-job consumption have positive and inhibitory effects, respectively, on the sustainable development of enterprises and the relationship between innovation efﬁciency and the sustainable development of enterprises. Compared to private enterprises, the positive effect of reasonable on-the-job consumption is smaller in state-owned enterprises, while the negative effect of excessive on-the-job consumption is larger in state-owned enterprises. Further research suggests that internal control can weaken the inhibitory effect of excessive on-the-job consumption on the positive relationship between the innovation efﬁciency and the sustainable development of enterprises, and the weakening effect is even greater after the implementation of anti-corruption policies. Anti-corruption policies and internal control can form a complementary force, effectively restrain the agency effect of excessive on-the-job consumption, and promote the sustainable development of enterprises. This research not only expands the research perspective of on-the-job consumption but also adds new theoretical explanations and empirical evidence for how to achieve the sustainable development of enterprises.


Introduction
How enterprises continuously survive and develop in a turbulent and fiercely competitive environment has become the focus of common concern in academia and industry. Schumpeter et al. stated that economic development is the result of innovation, and the development of any enterprise is inseparable from innovation [1]. In China, excellent enterprises represented by Huawei, Haier, Google, and ZTE have continued to grow through scientific and technological innovation and become internationally renowned enterprises with strong competitiveness, which has confirmed this view. At the same time, a large proportion of similar enterprises have also actively implemented the scientific and technological innovation strategy, but they have been eliminated in the fierce market competition. The contrast is so great that we cannot help thinking about the reasons for it. At present, research on the sustainable development of enterprises mainly focuses on the evaluation of sustainable development, and there has been little literature that has empirically demonstrated its antecedent variables or driving mechanism [2,3]. efficiency and sustainable development in order to provide some theoretical reference for the government to formulate relevant policies and enterprises to achieve sustainable development.
The possible contributions of this paper are as follows. First, the paper is focused on the research of factors affecting the sustainable development of enterprises. Previous research was mostly focused on the relationship between on-the-job consumption and the current performance of the enterprise, ignoring the impact on the sustainable development ability of enterprises. Second, we modified the previous measurement model of excessive on-thejob consumption. We adopted a more accurate and reasonable model to thoroughly and systematically consider the mechanism of reasonable and excessive on-the-job consumption on the sustainable development of enterprises. Third, this paper deepens the understanding of the relationship between enterprise innovation and sustainable development. From both subjective and objective perspectives, we explore the impact of on-the-job consumption with two economic natures and innovation efficiency on sustainable development, thus providing a new perspective for the study of the antecedent variables of the sustainable development of enterprises.

Reasonable On-the-Job Consumption and Excessive On-the-Job Consumption
The economic structure of on-the-job consumption is complex, including multiple economic components such as monetary compensation, normal duty consumption, and selfentertainment consumption [17]. There are obvious differences in the economic effects of the on-the-job consumption of different components. On-the-job consumption is naturally hidden, and the financial reports of listed companies rarely disclose this information, so it is difficult to separate the on-the-job consumption of different economic components from the total amount. Chen and Sun aggregated the on-the-job consumption of the same dominant component to a specific sample group through sample segmentation; based on this, they grouped and demonstrated the economic effects of on-the-job consumption of different components [18]. However, we believe that although this grouping measurement method can guarantee the objectivity of results, there may be selective bias due to herd classification, and the on-the-job consumption data of each company cannot be accurately obtained, which makes research conclusions unrepresentative.
Scholars generally believe that the key to distinguish various economic components from on-the-job consumption lies in "whether it is reasonable". Reasonable on-the-job consumption is the normal or necessary expenditure required to meet the operation and development aims of enterprises. Its goal is to promote the development of enterprises. It includes not only the on-the-job consumption that reasonably exists as a supplementary component due to the lack of explicit incentives for executives as an alternative system but also the normal on-the-job consumption that exists to maintain government-enterprise relationships and business partnerships. Controlling on-the-job consumption within a reasonable range can mobilize managers' enthusiasm, improve employee productivity, reduce employee turnover rate, establish friendly relations with government officials or business partners, and improve enterprise performance [19,20]. The difference between the total on-the-job consumption and the reasonable on-the-job consumption is the excessive on-the-job consumption, which has two economic properties of non-productivity and nonefficiency, including the self-entertainment consumption of executives seeking personal interests due to excessive self-motivation and expenditures used to please or cultivate the relationship with their superiors. The financial cost of excessive on-the-job consumption that has not been effectively controlled exceeds the performance increment it brings, which easily leads to corruption. Excessive on-the-job consumption is an expenditure that shareholders do not want to see and needs to be governed or eliminated [21][22][23]. The impact mechanism of on-the-job consumption on the sustainable development of enterprises is shown in Figure 1.
consumption is an expenditure that shareholders do not want to see and needs to be governed or eliminated [21][22][23]. The impact mechanism of on-the-job consumption on the sustainable development of enterprises is shown in Figure 1. Figure 1. The impact mechanism of on-the-job consumption and innovation efficiency on sustainable development.

On-the-Job Consumption and Sustainable Development
The definition of the sustainable development of enterprises has two levels. The broad level refers to the idea that enterprises can integrate internal and external resources for a long time, optimize resource allocation, adapt to the external environment, and develop in an all-round and coordinated manner with the economy, nature, and society; the narrow level refers to the long-term and healthy growth of the enterprise itself [24]. In this paper, we study how enterprises realize sustainable development from a narrow perspective. Supplementary on-the-job consumption can alleviate the double bias to a certain extent [25]. Firstly, as an alternative means of insufficient monetary salary incentives, on-the-job consumption can alleviate salary rigidity, compensate for a lack of explicit salary material incentives, improve work efficiency, and increase the future return on assets and enterprise value [10,16]. Rajan and Wulf found that CEOs' purpose when using special planes is to save time and improve work efficiency [19]. Secondly, according to status economics theory and Maslow's hierarchy of needs theory, supplementary on-the-job consumption is an easier way to express the relative social status, identity, and prestige of managers; help executives gain recognition and respect from relevant personnel inside and outside the enterprise; obtain spiritual satisfaction; and play a strong non-material incentive role for executives [26][27][28][29], thus encouraging executives to continuously improve their business operation and development ability. Rajan and Wulf also believed that providing executives with on-the-job consumption facilities such as special cars can improve their reputation and enable their instructions to be effectively implemented, and its incentive effect is significantly greater than that of monetary compensation of the same value [19]. In addition, supplementary on-the-job consumption with both scarcity and observability makes it easier to express employers' concern for employees and their recognition and appreciation of employees' performance, increase employees' sense of dependence on the enterprise, and improve employees' loyalty [27] so as to convey friendly messages inside and outside the enterprise and create a good atmosphere [30].
As an emerging and transition economy, China has a serious lack of a formal systems. In the context of Chinese culture, relationship capital is a special economic resource of companies. The normal job consumption used to maintain this relationship is the product of this social environment. Enterprises can establish a relationship network with the government and business partners through normal on-the-job consumption and enjoy more government and social resources, such as bypassing unreasonable government regulation, enjoying government subsidies and reducing taxes and fees [31], and obtaining more business orders, bank loans, and debt financing (among others) [12,32,33] in order to improve the efficiency of enterprises, reduce transaction costs, and bring sustainable economic benefits to enterprises.

On-the-Job Consumption and Sustainable Development
The definition of the sustainable development of enterprises has two levels. The broad level refers to the idea that enterprises can integrate internal and external resources for a long time, optimize resource allocation, adapt to the external environment, and develop in an all-round and coordinated manner with the economy, nature, and society; the narrow level refers to the long-term and healthy growth of the enterprise itself [24]. In this paper, we study how enterprises realize sustainable development from a narrow perspective. Supplementary on-the-job consumption can alleviate the double bias to a certain extent [25]. Firstly, as an alternative means of insufficient monetary salary incentives, on-the-job consumption can alleviate salary rigidity, compensate for a lack of explicit salary material incentives, improve work efficiency, and increase the future return on assets and enterprise value [10,16]. Rajan and Wulf found that CEOs' purpose when using special planes is to save time and improve work efficiency [19]. Secondly, according to status economics theory and Maslow's hierarchy of needs theory, supplementary on-the-job consumption is an easier way to express the relative social status, identity, and prestige of managers; help executives gain recognition and respect from relevant personnel inside and outside the enterprise; obtain spiritual satisfaction; and play a strong non-material incentive role for executives [26][27][28][29], thus encouraging executives to continuously improve their business operation and development ability. Rajan and Wulf also believed that providing executives with on-the-job consumption facilities such as special cars can improve their reputation and enable their instructions to be effectively implemented, and its incentive effect is significantly greater than that of monetary compensation of the same value [19]. In addition, supplementary on-the-job consumption with both scarcity and observability makes it easier to express employers' concern for employees and their recognition and appreciation of employees' performance, increase employees' sense of dependence on the enterprise, and improve employees' loyalty [27] so as to convey friendly messages inside and outside the enterprise and create a good atmosphere [30].
As an emerging and transition economy, China has a serious lack of a formal systems. In the context of Chinese culture, relationship capital is a special economic resource of companies. The normal job consumption used to maintain this relationship is the product of this social environment. Enterprises can establish a relationship network with the government and business partners through normal on-the-job consumption and enjoy more government and social resources, such as bypassing unreasonable government regulation, enjoying government subsidies and reducing taxes and fees [31], and obtaining more business orders, bank loans, and debt financing (among others) [12,32,33] in order to improve the efficiency of enterprises, reduce transaction costs, and bring sustainable economic benefits to enterprises.
Although reasonable on-the-job consumption increases enterprise expenses, it is necessary to maintain enterprise operations because this cannot only stimulate the executives and encourage them to pursue long-term interests but also create a good environment for the enterprise and obtain more external resources. In the long run, reasonable on-the-job consumption can reduce enterprise operating costs, improve operating efficiency, increase enterprise market value, stabilize enterprise core competitiveness, and radiate strong vital-ity [20], thus forming a virtuous circle of enterprise development and achieving sustainable profitability.
Excessive on-the-job consumption caused by information asymmetry and agency conflict is an opportunistic behavior. Its essence is rent-seeking behavior in which managers consume enterprise resources to increase personal utility under self-interested motives, requiring less or no effort. Managers use limited resources for excessive on-the-job consumption such as self-enjoyment and flattering superior officials, which inevitably causes a waste of resources, leads to high agency costs, reduces the enterprise's performance, and harms the interests of shareholders [22,23], none of which are conducive to the sustainable operation and development of an enterprise. A large number of studies have attributed excessive on-the-job consumption to hidden corruption and supported this view by demonstrating that the excessive on-the-job consumption is negatively correlated with earnings management, operational efficiency, and company performance [15,34].

Hypothesis 1 (H1):
Under the control of other factors, there is a positive correlation between reasonable on-the-job consumption and the sustainable development of enterprises.

Hypothesis 2 (H2):
Under the control of other factors, there is a negative correlation between excessive on-the-job consumption and the sustainable development of enterprises.

The Interaction between On-the-Job Consumption and Innovation Efficiency
Enterprise innovation is a special management behavior. From the perspective of management efficiency, its efficiency is expressed as the optimization degree of resource allocation in the innovation process, that is, the enterprise allocates innovation investment among different innovation projects; if the same amount of innovation investment can bring greater output through allocation, the optimization degree of innovation allocation is higher. Therefore, the higher the optimization degree of innovation resource allocation, the higher the innovation efficiency.
According to the risk compensation theory, supplementary on-the-job consumption can compensate for the cost of failure caused by managers' investment, weaken managers' tendency to over-avoid risks, and give them the motivation to consider various investments closely related to the long-term development of the enterprise. For example, Liu found that the lower the intensity of on-the-job consumption, the lower the executive's operational effort and the lower the innovation enthusiasm and innovation investment [13]. As a "status commodity" recognized by management, supplementary on-the-job consumption can help strengthen the management's sense of belonging, guide their behavior towards projects conducive to the long-term development of the enterprise (such as enterprise innovation), actively improve the original production process, strive to optimize the degree of resource allocation, continuously reduce production costs, and improve production efficiency to continuously improve the development capabilities and competitive advantages of the enterprise.
Implicit incentives mainly include political promotion, on-the-job consumption, and rent related to investment expansion [35]. For state-owned enterprises, it is difficult to achieve political promotion in the short term, and private enterprises cannot even talk about political promotion. Although reasonable on-the-job consumption also squeezes out productive resources conducive to enterprise development, such as innovation resources, the squeezed-out resources are relatively small compared to investment expansion. When the on-the-job consumption of supplementary components increases, managers' desire to obtain implicit incentives through enterprise investment expansion weakens and enterprise overinvestment accordingly decreases. Therefore, the on-the-job consumption of supplementary components can, to a certain extent, secure more sources of funds conducive to the long-term development of enterprises, improve the rationality of fund utilization, and promote the sustainable development of enterprises.
China's innovation support policies have not yet been perfected. Government officials have greater control and decision-making space in the company's enjoyment of innovation subsidies and other policies [36]. Rational enterprise executives often establish a relationship network with the government through normal on-the-job consumption and can basically realize that the innovation subsidies they obtain are greater than the squeezed innovation resources [11], reduce the cost of the social resources they obtain, and improve resource utilization in order to effectively improve the development ability of enterprises.
Due to information asymmetry, executives in an information-dominant position often take advantage of their positions to obtain excessive on-the-job consumption in the form of consumption and enjoyment. On the one hand, executives who enjoy excessive on-the-job consumption may be more immersed in a peaceful life, thus losing their willingness to invest in high-input, high-risk, and long-cycle innovative projects and long-term operational production [37], thus reducing the rationality of innovation investment and the motivation to pursue the long-term development of enterprises. On the other hand, excessive on-the-job consumption may lead to the rise in agency costs, inhibit innovationfinancing capabilities, squeeze enterprise production costs, reduce the optimization of resource allocation, and weaken the promotion effect of enterprise innovation on sustainable development.

Hypothesis 3 (H3):
Reasonable on-the-job consumption can strengthen the promotion of innovation efficiency on the sustainable development of enterprises.

Hypothesis 4 (H4):
Excessive on-the-job consumption may weaken the promotion of innovation efficiency in the sustainable development of enterprises.

Sample Selection and Data Sources
We selected China's A-share listed companies (companies limited by shares that issue ordinary shares in RMB and can be listed and traded on the Chinese stock market) from 2008 to 2019, excluding financial and insurance, real estate, and ST and *ST listed companies (ST and *ST represent listed companies with losses for two consecutive years and three consecutive years, respectively). In order to eliminate the influence of extreme values, the continuous variables were Winsorized at the 1% and 99% quantiles, and finally, 19,021 sample data of 2803 listed companies were obtained. In the above-mentioned sample data, the data related to patent situation came from the CIRD (China Innovation Patent Research Database, which contains data about patent application and authorization); the data on the establishment of various committees came from the CCER database (China Center for Economic Research database, including data on the capital market, money market, and macro economy); the basic status data of the company came from the CSMAR database (China Stock Market Accounting Research database, including the company, character characteristics, green economy, and other relevant data); and other data came from the WIND database (WIND information finance database, including stocks, funds, bonds, macroeconomics, financial news, and other relevant data).

Variable Definition
Sustainable development (Sgr): Sustainable growth rate (Sgr) refers to the maximum growth rate that an enterprise's sales can achieve without completely exhausting existing financial resources. This growth rate reflects the enterprise's operating performance through financial data, which can effectively measure the enterprise's long-term profitability and lasting competitive advantage, and it can be used to judge whether the enterprise's operating performance and financial status have achieved sustainable development [38,39]. Therefore, we chose Higgins' sustainable growth model to measure the sustainable development level of listed companies in China: where P is the net interest rate on sales, A is the asset turnover rate, T is the equity multiplier, and R is the retained earnings rate. Innovation efficiency (Score): this was measured by the optimization degree of innovation investment allocation among different innovation projects. The most commonly used efficiency evaluation methods are data envelopment analysis (DEA) and stochastic frontier analysis (SFA), both of which have their own advantages and disadvantages. Because the DEA model does not need to set parameters, we used the variable return to scale model (BBC) in the DEA model to estimate the optimization degree of innovation allocation. Enterprise innovation investment was selected as the input variable. Additionally, in order to better illustrate the output level of core technological innovation, the number of invention patent applications was selected as the output variable [40].
Excessive on-the-job consumption (Eper): Using the empirical estimation model proposed by Baobao Li and Shouchang Huang to quantify on-the-job consumption [41], the fixed management expenses were first removed from the total management expenses to obtain variable management expenses. The following model was regressed: where Mper 1 is the management expenses and Sale is sales revenue. The above model was regressed based on time series data, and the obtained intercept item was the fixed management expenses. The fixed management expenses were removed from the total management expenses, and the rest was the variable management expenses (Mper 2 ). Excluding the normal business operating expenses from the variable management expenses, the abnormal management expenses were obtained. The model was regressed as follows: where Size is the total assets; M/B is the price-to-book ratio; Mags is the number of managers, calculated logarithmically; Empl is the number of employees, calculated logarithmically; Meet is the sum of the number of meetings of the board of directors, the board of supervisors, and the shareholders, calculated logarithmically; Commt is the cumulative sum of 4 dummy variables (whether to establish audit committee, compensation committee, strategy committee, and nomination committee); and Locat is a regional factor to control the demand for business expenses caused by regional differences. If the registered place was in Beijing, Shanghai, Guangzhou, or Shenzhen, the value was 1; otherwise, it was 0. The above model was regressed by industry and year, and the residual vector after regression was the abnormal management expense rate, which represented the proportion of variable management expenses in income after excluding the expenses of normal business activities and could be used as the proxy variable of on-the-job consumption.
Zhang Yueming and Wu Chunlei believed that the residual term obtained by deducting the part of alternative explicit monetary compensation from on-the-job consumption is excessive on-the-job consumption, and their designed model did not estimate the on-the-job consumption used to maintain the relationship between enterprises, the government, and other businesses [42]. In view of this, we proposed a modified model, and the residual term obtained after the annual regression by industry was the excessive on-the-job consumption: where Mper is on-the-job consumption, Pay is the monetary salary of executives, Shary is the shareholding ratio of executives, Rel1 is the resource of government-enterprise relations (which is expressed as the ratio of the number of executives with government background to the total number of executive team), and Rel2 is the resource of business partnership (which is expressed as the ratio of the number of executives with part-time behavior to the total number of executive team). Specific variable definitions are shown in Table 1.

Model Setting
In order to test the impact of on-the-job consumption on sustainable development, models (5) and (6) were constructed: In order to test the mechanism of on-the-job consumption on the relationship between innovation efficiency and sustainable development, models (7) and (8) were constructed: In addition to the above research, we also examined the heterogeneity of the nature of property rights and the governance effects of internal control and anti-corruption in order to form the research framework diagram shown in Figure 2. In order to test the mechanism of on-the-job consumption on the relationship between innovation efficiency and sustainable development, models (7) and (8) were constructed: Sgri,t = α0 +α1Eperi,t + α2Scorei,t + α3Eperi,t× Scorei,t + α4∑βicontroli,t + εi,t In addition to the above research, we also examined the heterogeneity of the nature of property rights and the governance effects of internal control and anti-corruption in order to form the research framework diagram shown in Figure 2.

Descriptive Statistical Analysis
The descriptive statistical results of the main variables are shown in Table 2. Reasonable and excessive on-the-job consumption were estimated based on industry averages. A positive sign indicates that the observed value was higher than the industry average, and a negative sign indicates the opposite. The minimum and maximum values of

Descriptive Statistical Analysis
The descriptive statistical results of the main variables are shown in Table 2. Reasonable and excessive on-the-job consumption were estimated based on industry averages. A positive sign indicates that the observed value was higher than the industry average, and a negative sign indicates the opposite. The minimum and maximum values of reasonable on-the-job consumption (Nper) were −123.033 and 171.17, respectively, and the standard deviation was 19.049, which indicates that the dispersion of reasonable on-the-job consumption in enterprises is relatively high. This is because the normal operating expenses included in reasonable on-the-job consumption are different due to different enterprise scales and operation and management practices. The maximum value of excessive onthe-job consumption (Eper) was 108.939, and the minimum value was −9.523, indicating that there is a large difference in excessive on-the-job consumption among enterprises; additionally, the median (5.346) was greater than the mean (5.014), that is, many enterprises were found have a high degree of excessive on-the-job consumption and need to be supervised and controlled. The sustainable growth rate (Sgr) of enterprises ranged from −0.919 to 0.982, with a large difference, and the median (0.336) was less than the mean (0.350), indicating that the sustainable growth rate of most enterprises is low. For the innovation efficiency (Score), the minimum value was 0, the maximum value was 0.979, and the standard deviation was 0.316, which means that the innovation efficiency of different companies in the sample varies greatly. The descriptive statistical results of other control variables are shown in Table 2.

Multiple Regression Analysis
From columns 1 and 4 of Table 3, it can be seen that the regression coefficient between normal on-the-job consumption (Nper) and sustainable growth rate (Sgr) was 0.005, and the regression coefficient between excessive on-the-job consumption (Eper) and sustainable growth rate (Sgr) was −0.015; these were both significant at the level of 1%, indicating that reasonable on-the-job consumption has a positive impact on the sustainable development of enterprises and excessive on-the-job consumption for the purpose of maximizing personal utility leads to the consumption of enterprise resources, which is not conducive to the sustainable development of enterprises. As such, H1 and H2 are confirmed. From columns 2 and 5, it can be seen that the regression coefficient between the innovation efficiency (Score) and the sustainable growth rate (Sgr) was positive and significant at the level of 1%, indicating that innovation efficiency is conducive to improving the sustainable growth rate and is an important source of the sustainable development of enterprises. Column 3 reports the empirical results of model (7). The regression coefficient between the intersection term of reasonable on-the-job consumption and innovation efficiency (Nper i,t × Score) and sustainable growth rate (Sgr) was found to be 0.011 and significantly positive at the level of 1%, indicating that the interaction between reasonable on-thejob consumption and innovation efficiency promotes a sustainable growth rate, that is, reasonable on-the-job consumption strengthens the promotion of enterprise innovation on sustainable development. Accordingly, hypothesis H3 is verified. Column 6 presents the results of tests on the impact of the regulatory role of excessive on-the-job consumption and innovation efficiency on the sustainable development of enterprises. The results show that the coefficient of regression between the intersection term of excessive on-the-job consumption and innovation efficiency (Eper × Score) and sustainable growth rate (Sgr) was −0.072 and significantly at the level of 5%, and the main effect of innovation efficiency on sustainable growth rate was also found to decrease from 0.183 to 0.112. These results confirm H4, which states that the existence of excessive on-the-job consumption weakens the promotion of innovation efficiency on the sustainable development of enterprises. Table 3. Executives' on-the-job consumption, innovation efficiency, and sustainable development.  Note: t-values are in parentheses; ***, **, and * represent two-tailed statistical significance the 1%, 5%, and 10% levels, respectively.

1.
Index substitution test of main variables: We changed the measurement index of the explained variable. In order to ensure the reliability of the research results, the sustainable growth rate (Sgrv) calculated by the Van Horne model was used as the dependent variable [38], and the robustness of the model was analyzed: where P is the net interest rate on sales, A is the asset turnover rate, T is the equity multiplier, and R is the retained rate of return. We also changed the measurement index of the explanatory variable. Based on the research of Zhai et al. [43], an empirical estimation model of on-the-job consumption was constructed, and regression tests were conducted by industry and year. The fitting value obtained by the regression of the model was on-the-job consumption, and the residual was excessive on-the-job consumption (which represents the abnormal part of the total on-the-job consumption that deviates from the reasonable level). The difference between the total on-the-job consumption and the abnormal on-the-job consumption is the reasonable on-the-job consumption.
We changed the measurement of the adjustment variable. Referring to Xu et al. [44] and Sun et al. [45], we selected the natural logarithm of R&D expenses as the input variable and the natural logarithm of the total number of patents applied as the output variable, and then we re-estimated the innovation efficiency by using data envelopment analysis (DEA).

2.
We changed the treatment of extreme values. In order to avoid the influence of extreme values on the conclusion, continuous variables were Winsorized at the 1% and 99% quantiles. As a robustness test, we further strengthened the treatment of outliers, that is, we Winsorized all continuous variables at the 2% and 98% quantile levels.

3.
Parallel panel data: The sample data of the listed companies used in the previous test and analysis were non-parallel panel data. In order to improve the reliability of the research conclusions, we further processed the samples as parallel panel data, obtained 10,121 annual observations of 1054 companies, and again carried out empirical analysis.

4.
Propensity score matching method: In order to eliminate the endogenous problems caused by self-selection, the propensity score matching method was used for testing. Firstly, according to the median of reasonable on-the-job consumption, the whole sample was divided into a treatment group (Nper was higher than the median array) and a control group (Nper was lower than or equal to the median array). Secondly, by using the reasonable on-the-job consumption (Nper) of the treatment group as the explained variable and the control variable in model (5) as the explanatory variable, the propensity score was estimated by logit regression. Furthermore, the nearest neighbor matching method was used for one-to-one matching without reversal. We successfully matched 3846 sample values, that is, 7692 observations met the common support hypothesis and were retained. Finally, the matched samples were tested and analyzed (the excessive on-the-job consumption was the same as above, and a total of 7874 observations were obtained). Table 4 reports the estimated results of the robustness test. It can be seen that in a series of robustness tests, the previous regression conclusions did not substantially change, and they were relatively robust and reliable.   Note: t-values are in parentheses; ***, **, and * represent two-tailed statistical significance the 1%, 5%, and 10% levels, respectively.

Contextual Effects Based on the Characteristics of Company Property Rights
State-owned enterprises shoulder government tasks. Under government intervention, they do not have to form a competitive advantage through enterprise innovation and other projects that are conducive to the sustainable development of enterprises. The promotion of state-owned enterprise executives is similar to the promotion of government officials. Economic performance is only a part of the post-promotion assessment. Stateowned enterprise executives are likely not to care about enterprise innovation and often make enterprise development decisions based on political factors, and the demand for supplementary on-the-job consumption is relatively weak. The inherent political relevance of state-owned enterprises makes it easier to obtain innovation subsidies (among others), and the demand for obtaining social resources through the on-the-job consumption of normal positions is relatively small [11]. Therefore, compared to private enterprises, reasonable on-the-job consumption and innovation efficiency in state-owned enterprises play smaller roles in promoting the sustainable development of enterprises. Due to the absence of managers in state-owned enterprises, management is subject to less relevant constraints, the supervision of enterprise executives is not in place, and their on-thejob consumption expenditures are more arbitrary. Private enterprises are responsible for their own profits and losses. In terms of risk control, private enterprises have more urgent needs. The assessment of economic performance is the main goal of managers, which makes the on-the-job consumption of private listed companies occur more within the context of specified incentives, and the objectives of managers are often consistent with enterprise interests, thereby weakening the negative effect of excessive on-the-job consumption on the sustainable development of enterprises. Therefore, compared to private enterprises, excessive on-the-job consumption in state-owned enterprises plays a greater role in weakening the sustainable development of enterprises.
The method of grouping according to the nature of property rights and establishing the regulatory effect of the nature of property rights was used here to verify the situational effect of the nature of property rights, that is, the impact of on-the-job consumption and innovation efficiency on the sustainable development of enterprises under different natures of property rights. The test results are shown in Table 5. From columns 1 and 2, it can be seen that whether in state-owned or private enterprises, the regression coefficients of the intersection term of reasonable on-the-job consumption and innovation efficiency (Nper × Score) were both positive and significant at the levels of 1% and 5%, respectively, which shows that reasonable on-the-job consumption is helpful in promoting sustainable development in both state-owned and private enterprises. Hypothesis H1 is accordingly verified again. From column 3, it can be seen that the regression coefficient of the intersection term of property right nature, reasonable on-the-job consumption, and innovation efficiency (Nper × Score × State) was significantly negative and significant at the level of 1%, which indicates that the situational effect of state-owned enterprises is small, that is, compared to private enterprises, the promotion effect of reasonable onthe-job consumption on the positive relationship between innovation efficiency and the sustainable development of enterprises in state-owned enterprises is weakened. From columns 4 and 5, it can be seen that whether in state-owned or private enterprises, the regression coefficients of the intersection term of excessive on-the-job consumption and innovation efficiency (Eper × Score) were negative and significant at the levels of 10% and 5%, respectively. Hypothesis H2 is verified again. It can be seen from column 6 that the regression coefficient of Eper × Score × State was significantly negative, as was the regression coefficient of Eper × Score, and it was significant at the 5% level, indicating that the situational effect of state-owned enterprises was greater. In other words, compared to private enterprises, excessive on-the-job consumption in state-owned enterprises plays a greater role in weakening the positive relationship between innovation efficiency and the sustainable development of enterprises. Due to the unproductiveness and inefficiency of excessive on-the-job consumption, the control of excessive on-the-job consumption behavior is particularly important in enterprise operation and management. As such, we further studied the governance mechanism of excessive on-the-job consumption. Internal control can play an important role in supervision, control, information transmission, and feedback in the process of corporate governance through a series of control measures such as budget processes, approval authority, expenditure standards, and supervision. Therefore, a sound internal control system is an important guarantee for a company's high-quality operation, as it can help enterprises straighten out the corporate governance structure, scientifically and reasonably allocate resources, prevent the waste and loss of resources, and achieve 'the best use of everything' in order to effectively curb the agency effect of excessive on-the-job consumption.
The internal control index of the Dibo database was selected as the measurement index (Ic) of the internal control, and the specific test results are shown in Table 6. It can be seen from column 1 that the coefficient of Ic × Eper × Score was significantly negative, as was the regression coefficient of Eper × Score, and it was significant at the 10% level. According to columns 3 and 4, in the high-quality internal control group, the regression coefficient of Eper × Score was −0.004 and significant at the 10% level, which was significantly smaller than that of the low-quality internal control group. The above results show that although internal control cannot completely eliminate the negative effects of excessive on-the-job consumption, the negative effects of excessive on-the-job consumption are smaller in enterprises with high-quality internal control, that is, compared to the group with lowquality internal control, the weakening effect of excessive on-the-job consumption on the positive relationship between innovation efficiency and the sustainable development of enterprises is smaller in a group with high-quality internal control. Table 6. Governance mechanism test of excessive on-the-job consumption.  Note: t-values are in parentheses; ***, **, and * represent two-tailed statistical significance the 1%, 5%, and 10% levels, respectively.

The Synergistic Governance Effect of Anti-Corruption Policies and Internal Control
In order to promote the research on the governance of enterprise internal control on the excessive on-the-job consumption agency effect, we further analyzed the synergistic governance effect of anti-corruption policies and internal controls. After the 18th National Congress of the Communist Party of China, the Political Bureau of the CPC Central Committee successively issued a series of documents, such as the "six prohibitions" and the anti-"four winds", and regulated and controlled the excessive on-the-job consumption of the officials and senior executives of state-owned enterprises through an investigation into and punishment of abuse of power, corruption and bribery, illegal misappropriation of state property, eating and drinking with public funds, etc., in key posts. On the one hand, this has increased the rental cost of government officials' on-the-job consumption, optimized the internal control environment and sustainable operation environment of enterprises, and increased the willingness of enterprises to innovate. On the other hand, anti-corruption can also effectively regulate on-the-job consumer spending, improve corporate governance, and promote the long-term development of enterprises by reducing the incompleteness of contracts, alleviating double agency conflicts, reducing insider control, and strengthening external governance constraints. In addition, in the anti-corruption context, the "sacking" of higher-level corrupt officials also opens the promotion channels of lower-level officials. In order to obtain political promotion opportunities, senior executives need to achieve relatively good performance, which is also conducive to promoting the sustainable development of enterprises.
Using the closing of the 18th CPC National Congress as the dividing point, the company data after the scenes have an anti-corruption background. If the value of the anticorruption background (Mc) was found to be 1, the year after anti-corruption (2013-2019) was indicated as having an anti-corruption background; if the value of Mc was 0, the year before anti-corruption (2008-2012) was indicated as not having an anti-corruption background. As shown in column 2 of Table 6, the coefficient of Mc × Eper × score was significantly negative. Columns 5 and 6 report the grouping test results regarding whether an enterprise has anti-corruption background. In the group with anti-corruption background, the regression coefficient of Ic × Eper × score was −0.002 and significant at the level of 10%, which was smaller than that of the group before anti-corruption, indicating that in the process of controlling excessive on-the-job consumption, the anticorruption policy can play a strong external governance role, standardize the business behavior of enterprise executives, and weaken the agency effect of excessive on-the-job consumption. Anti-corruption policies and internal controls can complement each other to form an institutional joint force, effectively curb the agency effect of excessive on-the-job consumption, and promote the sustainable development of enterprises.

Summary and Conclusions
Based on the sample data of A-share listed companies from 2008 to 2019, we constructed a theoretical model between on-the-job consumption, innovation efficiency, and the sustainable development of enterprises from the perspective of the synergy of control and innovation. We have also empirically demonstrated the impact of reasonable and excessive on-the-job consumption on the sustainable development of enterprises and the relationship between innovation efficiency and the sustainable development of enterprises. The results showed that reasonable on-the-job consumption has a positive effect on the sustainable development of enterprises and the positive relationship between innovation efficiency and the sustainable development of enterprises, while excessive on-the-job consumption has a negative effect on the sustainable development of enterprises and the positive relationship between innovation efficiency and the sustainable development of enterprises. The heterogeneity test revealed that compared to private enterprises, reasonable on-the-job consumption in state-owned enterprises has a smaller effect on promoting the positive relationship between innovation efficiency and the sustainable development of enterprises, and excessive on-the-job consumption has a larger effect on weakening the positive relationship between them. Further research found that internal control can suppress the agency effect of excessive on-the-job consumption and weaken the inhibitory effect of excessive on-the-job consumption on the positive relationship between innovation efficiency and the sustainable development of enterprises. We found that after the implementation of anti-corruption policy, the situational effect of internal control is greater, that is, in the group with an anti-corruption background, the weakening of internal control was shown to have a greater inhibitory effect on excessive on-the-job consumption.
The research conclusions of this paper have certain theoretical guiding significance for enterprises to formulate sustainable development strategic plans. Based on theoretical analysis and empirical evidence, the suggestions are as follows:

1.
Equal attention should be paid to control and innovation. In the process of the sustainable development of enterprises, enterprises should not only actively improve innovation efficiency but also balance the symbiotic relationship between control and innovation, actively improve corporate governance structure, reasonably standardize on-the-job consumption, reduce power rent-seeking behavior as much as possible, and make resource allocation decisions conducive to the sustainable development of enterprises. On the one hand, enterprises need to provide managers with appropriate supplementary on-the-job consumption to improve their work efficiency. On the other hand, considering the relationship resource as a special, heterogeneous, and scarce resource, managers should be given appropriate authority so that they can use onthe-job consumption to maintain the network relationship between the government and companies, as well as effectively promote the development of enterprises.

2.
On-the-job consumption should be incorporated into corporate governance. At present, Chinese listed companies only disclose the total amount of executive currency compensation; its specific composition and other on-the-job consumption information are not required to be disclosed, and on-the-job consumption is out of the sight of corporate governance. Accordingly, on the one hand, enterprises should actively learn from the practice of the US SEC and disclose not only the total amount of executive compensation but also the specific composition of its salary incentive, including on-the-job consumption. On the other hand, enterprises should strengthen internal financial control and audit supervision, incorporate on-the-job consumption into the corporate governance framework, actively consider the restrictive role of internal control on the on-the-job consumption behavior of the management, restrain the rent-seeking behavior of senior executives using their own rights to obtain excessive on-the-job consumption at the source, and prevent the phenomenon of digesting excessive on-the-job consumption into normal operating expenses. 3.
The role of government control should be accurate. In the short term, the government strengthens the systematic supervision of corporate monetary compensation and on-the-job consumption, which can, to a certain extent, prevent senior executives from using job consumption to facilitate self-interested behavior and reduce agency costs. However, in essence, government intervention is an important motivation for excessive on-the-job consumption. Government departments should constantly improve the legal system, strive to break the government's monopoly on social resources, reduce excessive intervention in the market, truly achieve the marketization of resource allocation, fundamentally reduce on-the-job consumption caused by rentseeking behavior, and compress the rent-seeking space from the source.

4.
Reasonable on-the-job consumption standards should be appropriate. Compared to British and American countries with sound formal systems, in China's localization environment, the positive effect of normal duty on-the-job consumption of Chinese listed companies is likely to be obtained by undermining the optimal allocation of market resources. This paper is the only empirical study that has found that reasonable on-the-job consumption can promote the sustainable development of Chinese companies. However, from the perspective of optimizing the allocation of social resources, we do not encourage listed companies to improve the level of normal onthe-job consumption; instead, they should promote the implementation of reasonable determination and strict norms. Managers should comprehensively consider how to realize effective incentive and reasonable access to social resources in order to comprehensively improve the innovation efficiency and sustainable development ability of enterprises.
Author Contributions: The authors contributed equally to all sections of this paper. All the authors contributed to the research design. P.W. prepared the first draft. All authors have read and agreed to the published version of the manuscript. Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the first author P. Wang.