The impact of risk preference of top management team on re-innovation behavior after innovation failure

Abstract

This paper takes the listed pharmaceutical manufacturing firms in China’s A-share market from 2011 to 2017 as the research sample, from the perspective of top managers’ pay gap, employs principal component analysis and general least square method to empirically investigate the effect of risk preference of top management team(TMT) on the re-innovation behavior after the failure of innovation. The study found that the risk preference of TMT is positively correlated with the re-innovation input and brand-new innovation after the failure of innovation, but not with the supplementary innovation. Besides, the pay gap not only has a positive moderating effect on the positive correlation between the risk preference of TMT and the re-innovation input, but also on the positive correlation between the risk preference of TMT and the brand-new innovation after the failure of innovation. The findings of this study contribute to: (1) through empirical research on the impact of TMT risk preference on re-innovation behavior after innovation failure, expand the relevant research content and research methods of TMT and innovation failure to make the research results more convincing; (2) by setting a reasonable executive compensation gap, TMT can avoid blindly choosing brand-new innovation behavior after innovation failure with the increase of risk preference innovation, ignoring the potential value of innovation failure projects without supplementary innovation, improving the re-innovation behavior of TMT after innovation failure and improving the re innovation success rate after innovation failure.

Keywords

TMT risk preference re-innovation after the failure brand-new innovation supplementary innovation pay gap

1 Introduction

According to the theory of organizational behavior, the feedback information of past innovation activities will affect the future decision-making behavior of top managers, and then affect the future innovation behavior of firms [5]. Generally, successful projects in the past will continue in the future, and failed projects in the past should be stopped [21]. When a firm fails in innovation, managers can either give up the failed innovation project and start a brand-new innovation, or adjust and modify the failed project to carry out supplementary innovation [15]. With the increasing complexity of the external environment, it is difficult for any outstanding manager to make rational and satisfactory innovation decisions in a rapidly changing environment with his own wisdom and experience alone. Therefore, as an important strategic decision-making resource in modern firms, the choice of innovation behavior of TMT is directly related to the success or failure of the firm in the future, and its risk perception, cognition and preference will affect their final decision-making. Because the types and degrees of risks faced by brand-new innovation and supplementary innovation activities are different, and the innovation decisions made by TMT are often determined by the subjective consciousness of team members, therefore, as an important psychological feature affecting decision-making of managers, risk preference can better reflect the attitude of managers to uncertainty [28]. The level of risk preference of managers will affect the innovation willingness and decision-making of firms, and then have a significant impact on the subsequent re-innovation behavior of firms [12]. When a firm fails in innovation, the decision-making process of its re-innovation behavior is often more complicated due to the influence of previous failure. TMT not only needs to fully investigate the internal and external environment, but also re-examine the attribution of innovation failure. After summarizing and learning from the failure experience, TMT makes the final choice of re-innovation behavior, and the difference of risk preference of different managers also has different influence on the re-innovation behavior after innovation failure. If the risk preference of TMT is the driver of re-innovation behavior after the failure of innovation, then the compensation gap is the fuel. This is because the incentive effect of the internal compensation gap will affect the willingness of TMT to take risks [13]. Thus the design of reasonable pay gap is not only conducive to reducing the adverse impact of innovation failure on the firm itself, but also to reducing the possibility that TMT becomes too conservative and cautious about risk due to failure, unwilling to invest in innovation, resulting in high agency costs. A reasonable pay gap is more conducive to the re-innovation behavior after innovation failure. It can be seen that when we study the relationship between the risk preference of TMT and re-innovation behavior after the failure of innovation, we should consider the role of the pay gap between them. Because the external factors such as market and institutional environment can only be predicted, firms can only adjust the risk preference structure of TMT and design reasonable pay gap to improve their innovation behavior after failure by adapting to the external environment.

At present, it was still lack of systematic research on the risk preference of TMT and firms failure re-innovation behavior in practice and theory. Relevant researches are mainly carried out from the following aspects. Firstly, the impact of TMT’s personal characteristics or risk preference on enterprise innovation [2]; Secondly, from the perspective of failure learning, analyze the relationship between innovation failure and re innovation through case or theory [14].Thirdly, the computer method is used to simulate the failure re-innovation [32] or to build a group decision-making model [9,10,33 , 9,10,33]. Finally, it analyzes the enterprise innovation from the external institutional environment [31]. But these aspects of research are worth further development. Firstly, from the perspective of the influence of the individual characteristics of the TMT or the risk preference on the firm innovation, the influence of the risk preference of the TMT on the firm innovation failure re-behavior and the the role of pay gap between TMT risk preference and firm innovation failure re-innovation behavior are not considered. Secondly, from the perspective of failure learning, there are great differences in the existing conclusions. One view is that failure learning can stimulate managers’ re-innovation behavior [6]; the other view is that negative emotions of failure will hinder managers from learning from failure, and have a negative impact on the re innovation behavior after failure [4]. Thirdly, the accuracy of computer simulation or group decision-making model is still questionable. Finally, although the research on enterprise innovation from the perspective of institutional environment is innovative to some extent, it ignores the most important internal factor of top management team. Therefore, the research space is left for this paper.

The innovation of this paper may be reflected in: (1) It expands the research on TMT and innovation failure. Unlike most previous studies, which study innovation input or performance from the perspective of individual basic characteristics or risk preference of TMT, this paper studies the impact of risk preference of TMT on re-innovation behavior after innovation failure, and expands the relevant research on TMT and innovation failure; (2) It expands the research perspective. From the perspective of executive compensation gap, this paper studies its role between the risk preference of the TMT and the re-innovation behavior of enterprise innovation failure.

The remaining structure of this paper is as follows. Section 2 is literature review and theoretical hypothesis. Research methods including data sources and main variables are introduced in Section 3.Section 4 is model building and the analysis of empirical results. Section 5 summarizes the conclusion of this paper.

2 Literature review and theoretical hypotheses

2.1 Impact of risk preference of TMT on the innovation behavior after the failure of innovation

Existing research show that the level of TMT risk preference depicts their psychological attitude towards uncertainty, and there are obvious differences in the attitudes of different managers towards risk [8]. Therefore, the risk preference of TMT shows the comprehensive performance of risk tendency of all top managers in the face of decision-making or uncertainty. It has more direct impact on important innovation strategic decision-making than background characteristics such as age and gender [27]. When there are more risk preference members in the team, the firms tend to pay more attention to the opportunities of reform or innovation. They will pay more attention to the transformation of R & D input into firm performance, and they are more keen to seize the market by increasing R & D human and material resources input and developing new products to achieve the innovation goals of the firm [23]. Such kind of risk-taking spirit and strategic decision-making ability, organization coordination and resource integration ability can promote effective transformation of R & D input to firm performance [22]. When there are more risk averse members in the team, the firm not only tends to give up input projects that have extra income and are beneficial to the shareholders, but also reduce the innovation activities of the firm, namely unwilling to invest too much in new product research and development and new market development. When studying medical manufacturing industry in the United States, Maslach D [5] divides the innovation behavior after the failure of firm innovation into brand-new innovation and supplementary innovation. He believes that brand-new innovation is one that firms carry out completely different from the existing innovation activities. And supplementary innovation is the supplementary input of firms in existing innovation activities. When a firm fails in innovation, the decision-making process of top managers is more complicated because of the influence of previous failure results. On the one hand, the attribution result of innovation failure will affect managers’ re-innovation behavior. The main basis for innovation failure firms to make re-innovation decisions is managers’ subjective attribution of failure. This differentiated attribution result of failure will cause innovation failure firms to overestimate the risk of re-innovation, and make managers pessimistic about the subsequent innovation of firms, so as to make managers pessimistic about the failure of firms Inhibition of re-innovation [25]. On the other hand, decision-making of re-innovation is based on re-judgment of the resource stock of previous failure results [1]. As a result, TMT of the failed firm should make in-depth consideration of the loss degree of technological innovation, its own innovation ability, re-innovation cost, innovation failure stage and other factors, so as to increase the uncertainty of re-innovation to a certain extent. Because the essence of firm innovation is the process of continuous re-innovation, the re-innovation behavior of firm innovation failure is the re-innovation input choice made by TMT after summary and learning of the causes and experiences of innovation failure and re-integration of failure resources. The re-innovation input after firm failure consists of brand-new innovation and supplementary innovation.

To sum up, managers’ risk preference can subjectively mobilize innovation enthusiasm of firms and effectively promote exploratory innovation activities of firms [20]. Based on these, this paper proposes the following assumption:

Hypothesis 1: The higher the risk preference of TMT is, the greater the re-innovation input is after innovation failure.

Hypothesis 1a: The higher the risk preference of TMT is, the greater the supplementary innovation is after innovation failure.

Hypothesis 1b: The higher the risk preference of TMT is, the greater the brand-new innovation is after innovation failure.

2.2 The moderating effect of pay gap

In existing studies, pay mainly includes monetary pay and non-monetary pay. The pay of top managers mainly consists of salary, bonus and stock option. However, most of the firms in our country do not implement the incentive policy of stock option, so scholars employ monetary income to measure the pay gap among top managers when they study the pay gap. At present, few scholars directly study the relationship between risk preference and innovation input of TMT by using the pay gap as moderating variable. The moderating effect of pay gap mainly focuses on the characteristics of top managers and R & D input or performance of firms. He Xia et al. [24] believed that the gap of top managers monetary pay has important moderator effect between the gender of TMT and R & D input, and pay gap has different moderator effects on the relationship between age, education level, tenure and other background characteristics of top managers and firm performance [16]. When studying the moderator effect of promotion incentive on the relationship between the characteristics of TMT and input efficiency, Li Huimin [7] found that monetary pay gap has the same moderator effect on the relationship between top managers’ age, education, tenure and input efficiency, but has reverse moderator effect on the relationship between gender and input efficiency. Xie Bo [3] chose GEM listed companies as the research object, and found that the monetary pay gap exerts strong inhibition effect on the relationship between the professional background heterogeneity of TMT and firm performance. As the personal characteristics of top managers are important indicators to measure their risk preference, the incentive effect brought by pay gap of top managers also has an important impact on their willingness to take risks [19]. Therefore, this paper believes that pay gap will play a moderator role between the risk preference of TMT and re-innovation behavior after the failure of innovation. Based on these, this paper proposes the following assumption:

Hypothesis 2: The pay gap of top managers has positive moderating effect on the relationship between risk preference of TMT and re-innovation input after the failure of innovation.

Hypothesis 2a: The pay gap of top managers has positive moderating effect on the relationship between risk preference of TMT and supplementary innovation after the failure of innovation.

Hypothesis 2b: The pay gap of top managers has positive moderating effect on the relationship between risk preference of TMT and brand-new innovation after the failure of innovation.

3 Methodology

3.1 Data

In the process of sustainable development, the time limit of firm strategic objectives is usually more than 5 years. According to the Guidelines for Classification of Listed Firms (Revised in 2012) issued by China’s Securities Moderator Commission, this study selects listed pharmaceutical manufacturing firms in China A-share market from 2011 to 2017 as samples. The sample data are obtained and screened as follows:

(1) According to the definition of the major variables and time interval of the sample, the missing data of relevant indicators and the sample of listed firms that do not meet the time interval of 2011–2017 are eliminated, excluding veterinary, agricultural drugs in the range of non-ADRs monitoring, and drug manufacturers are further matched manually by using the information disclosed in the official websites of listed firms and annual reports of firms. A total of 398 subsidiary drug manufacturers included by listed pharmaceutical manufacturing firms are obtained. On this basis, combined with adverse drug reaction database of pharmaceutical data center on national population and health science data sharing platform and the national adverse drug reaction monitoring system, information retrieval is carried out with keywords of “manufacturer” and the name of the drugs produced by the listed firms. When the ADRs produced by a listed firm’s affiliated firms occur in a certain year, that is, the sample firms failed in innovation in that year and the firm is an innovation failure firm.

(2) In addition to the data to be sorted out manually, other relevant data mainly come from CSMAR and wind databases.

According to the above principles, this paper finally obtained 228 effective research samples from 38 listed pharmaceutical manufacturing firms in 2011–2017, all of which had adverse drug reactions in 2011 and these firms have developed adverse drug reactions one after another in 2012–2017. The data of risk preference and pay gap of TMT is limited to 2011–2016, and the data of failure re-innovation behavior is limited to 2012–2017. The reason for this treatment is that since the innovation input of a firm is a strategic input, when a firm innovation failure happened in 2011, TMT will make a strategic plan for the re-innovation behavior of the firm from 2012 to 2017, or even longer.

3.2 Variable definition

3.2.1 Dependent variable

Re-innovation input after the failure of innovation (reinnovation) & supplementary innovation(incremental) & brand-new innovation (novel): Domestic and foreign firms mainly use the ratio of R & D expenditure to total assets, the ratio of R & D expenditure to main business income, and the natural logarithm of R & D expenditure to measure innovation input. According to the current accounting standards for Business Firms No. 06 - intangible assets, R & D expenditure is divided into research stage expenditure and development stage expenditure. Among them, research stage expenditure is included in the expensed R & D expenditure of the current period in the notes to the income statement. It is to acquire and understand new scientific and technological knowledge. In fact, expenditure of this stage is exploratory and could be viewed as brand-new innovation. Development stage expenditure is included in the capitalized R & D expenditure of the current period in the notes to the income statement. It is the improvement or application of research results obtained in the research stage or externally. Actually, the expenditure at this stage is a kind of the supplementary innovation investment. Therefore, combined with the definition of Maslach D [5], the brand-new innovation matches research stage expenditure on accounting recognition, and supplementary innovation matches development stage expenditure on accounting recognition. The failed re-innovation input is the sum of brand-new innovation and supplementary innovation. It is the re-innovation input of the firm based on the previous failed technology results, measured by the natural logarithm of the total R & D expenditure of the failed firm in the next year. Supplementary innovation and brand-new innovation refers to the supplementary innovation and brand-new innovation of firms based on the previous failed technical results, measured by the natural logarithm of the capitalized R & D expenditure and the expensed R & D expenditure of failed firms in the next year respectively.

3.2.2 Independent variable

Risk preference of TMT (mrp): In the existing literature, the measurement of risk preference of TMT is not uniform [11]. In view of the special situation of China’s capital market, this paper chooses to use the method of Cain et al. [18] to build TMT risk preference evaluation index system from firm to individual. Time series principal component analysis is used to calculate the risk preference score of TMT as the value of this variable.

On the basis of the research of the predecessors’ [26], in combination with the actual situation in China, this paper carries out forward processing and standardization processing on the risk preference evaluation indexes of TMT (Table 1), as well as KMO test and Bartlett’s spherical test to ensure that the selected indexes meet the basic requirements of time series principal component analysis, and then establishes the risk preference evaluation model of TMT: $mrp = \sum λ_{i} * f_{i}$ (1)

Table 1

Risk preference evaluation index system of TMT

Dimension	Name	Definition
Firm level	Quick ratio	Quick assets/current liabilities
	Ratio of short-term risk assets	(trading financial assets+accounts receivable)/current assets
	Current liability ratio	Current liabilities/total liabilities
	Asset-liability ratio	Total liabilities/total assets
	Long term asset suitability rate	(total owner’s equity+total long-term liabilities)/(total fixed assets+total long-term investments) ^*100%
	Core earnings ratio	Operating income/(operating income+fair value change profit and loss+investment income+non-operating income)
	Cash ratio	(cash+trading financial assets)/current liabilities
	Rate of long-term use of short-term funds	1- (non-current liabilities+shareholders’ equity)/non-current assets
Individual level	The proportion of male executives	The dummy variable, male 1, female 0. Proportion of male executives to total executives
	Average age of top management team	Average of the sum of top executives’ ages
	Average education level of top management team	Technical secondary school and below = 1, junior college = 2, bachelor = 3, master = 4, master above = 5;Average of the sum of executive education
	Average tenure of top management team	Average of the total number of months of executive cumulative tenure
	Average total wealth of top management team	The natural logarithm of the average value of the sum of stock market value and annual salary held by top executives at the end of the year
	Average number of shares held by top management team at the end of the year	The natural logarithm of the average number of shares held by executives at the end of the year
	Average risk assets ratio of top management team	Average ratio of year-end stock market value to annual salary and year-end stock market value of executives

Note: “Dimension” indicates that principal component indicators are selected at the individual and enterprise levels;”Name: is the indicator name;”Definition” represents the calculation method of each index.

Among them, mrp represents the risk preference score of TMT; λ_i indicates the variance contribution rate corresponding to each principal component; f_i represents the score of each principal component. Six principal components were extracted from 15 indicators, the scores of each principal component and the weight of corresponding variance contribution rate were substituted into model (1), and the final score was calculated as the assessment value of the risk preference of TMT in this paper.

3.2.3 Moderating variable

Pay gap(gap): As for the measurement of top managers’ pay gap, scholars at home and abroad mainly adopt absolute pay gap method, relative pay gap method, Gini coefficient method and pay gap variation coefficient method. In this paper, the relative pay gap method of Zhu [17] is selected for calculation.

3.2.4 Control variable

Based on existing literature [2,5,29 , 2,5,29], this paper controls firm scale (size), Proportion of independent directors (dlds), firm growth (growth), nature of firm ownership (state), returns on assets (roe) and other variables. The definitions of major variables are shown in Table 2.

Table 2
The definitions of major variables

Type Name Symbol Definition

Dependent variable Re-innovation input after the failure of innovation Re-innovation The failed re-innovation input is the sum of brand-new innovation and supplementary innovation. It is the re-innovation input of the firm based on previous failed technology results, measured by the natural logarithm of the total R &D expenditure of the failed firm in the next year

Supplementary innovation Incremental Supplementary innovation refers to the supplementary innovation of firms based on the previous failed technical results, measured by the natural logarithm of the capitalized R &D expenditure of failed firms in the next year

Brand-new innovation Novel Brand-new innovation refers to the brand-new innovation of firms based on the previous failed technical results, measured by the natural logarithm of the expensed R &D expenditure of failed firms in the next year

Independent variable Risk preference of TMT Mrp Results of time series principal component analysis

Moderator Variable Pay gap Gap The ratio of the average salary of the top three managers to the average salary of the rest top managers

Control variables Firm scale Size The natural logarithm of total assets

Proportion of independent directors Dlds Number of independent directors / board of directors

Firm growth Growth The growth rate of main business income, (main business income of this year – main business income at the beginning of this year) / main business income at the beginning of this year

Nature of firm ownership State Virtual variable, state = 1 for state-owned firms, state = 0 for non-state-owned firms

Returns on assets Roe Net profit/average total assets, average total assets = total end-of-period balance of assets

Type	Name	Symbol	Definition
Dependent variable	Re-innovation input after the failure of innovation	Re-innovation	The failed re-innovation input is the sum of brand-new innovation and supplementary innovation. It is the re-innovation input of the firm based on previous failed technology results, measured by the natural logarithm of the total R &D expenditure of the failed firm in the next year
	Supplementary innovation	Incremental	Supplementary innovation refers to the supplementary innovation of firms based on the previous failed technical results, measured by the natural logarithm of the capitalized R &D expenditure of failed firms in the next year
	Brand-new innovation	Novel	Brand-new innovation refers to the brand-new innovation of firms based on the previous failed technical results, measured by the natural logarithm of the expensed R &D expenditure of failed firms in the next year
Independent variable	Risk preference of TMT	Mrp	Results of time series principal component analysis
Moderator Variable	Pay gap	Gap	The ratio of the average salary of the top three managers to the average salary of the rest top managers
Control variables	Firm scale	Size	The natural logarithm of total assets
	Proportion of independent directors	Dlds	Number of independent directors / board of directors
	Firm growth	Growth	The growth rate of main business income, (main business income of this year – main business income at the beginning of this year) / main business income at the beginning of this year
	Nature of firm ownership	State	Virtual variable, state = 1 for state-owned firms, state = 0 for non-state-owned firms
	Returns on assets	Roe	Net profit/average total assets, average total assets = total end-of-period balance of assets

Note: The data are collated by the authors.

4 Results analysis

4.1 Hypothesis test

According to the above theoretical analysis and literature review, the model of the relationship between the risk preference of TMT and the re-innovation behavior after innovation failure may be a moderated model. Therefore, according to Wen et al. [30] on the test process of moderator effect, this paper first tests the impact of risk preference of TMT on the re-innovation behavior after innovation failure (hypothesis 1, 1a, 1b). The models to be tested are as follows:

$\begin{matrix} {reinnovation}_{(i, t + 1)} = α + β_{1} * {mrp}_{(i, t)} + β_{2} * {size}_{(i, t)} \\ + β_{3} * {dlds}_{(i, t)} + β_{4} * {growth}_{(i, t)} \\ + β_{5} * {roe}_{(i, t)} + β_{6} * {state}_{(i, t)} + μ_{(i, t)} \end{matrix}$ (2) $\begin{matrix} {incremental}_{(i, t + 1)} = α + β_{1} * {mrp}_{(i, t)} + β_{2} * {size}_{(i, t)} \\ + β_{3} * {dlds}_{(i, t)} + β_{4} * {growth}_{(i, t)} \\ + β_{5} * {roe}_{(i, t)} + β_{6} * {state}_{(i, t)} + μ_{(i, t)} \end{matrix}$ (3) $\begin{matrix} {novel}_{(i, t + 1)} = α + β_{1} * {mrp}_{(i, t)} + β_{2} * {size}_{(i, t)} \\ + β_{3} * {dlds}_{(i, t)} + β_{4} * {growth}_{(i, t)} \\ + β_{5} * {roe}_{(i, t)} + β_{6} * {state}_{(i, t)} + μ_{(i, t)} \end{matrix}$ (4)

Among them, reinnovation_(i,t+1), novel_(i,t+1), incremental_(i,t+1) represent the i-th sample firm in period t + 1’s re-innovation input after the failure of innovation, brand-new innovation, respectively,as well as the supplementary innovation. mrp_(i,t) indicates the risk preference of TMT of the i-th sample firm in t period. size_(i,t), dlds_(i,t), growth_(i,t), roe_(i,t), state_(i,t) represent the i-th sample firm in period t’s firm scale, proportion of independent directors, firm growth, nature of firm ownership respectively, as well as the returns on assets. β₁ to β₆ are regression coefficients, indicating the influence degree of independent variable and control variable on dependent variable. α is constant term, μ_(i,t) is random disturbance term.

According to a series of results shown in Table 3, the F value of model (2) is 55.201 with the significance level P < 0.01; the model (2) passes the F test. The risk preference of TMT (taken as the independent variable) has a significant positive effect on the re-innovation inputs on the basis of the failure of firm innovation. The standardized regression coefficient is 0.107 (P < 0.01), which indicates that the higher the risk preference of TMT, the greater the re-innovation inputs are after the failure of innovation. So the hypothesis 1 has been verified. As for the model (3), the F value is 1.403 with the significance level P > 0.1, and it doesn’t pass the F test. The risk preference of TMT (taken as the independent variable) doesn’t have a significant positive effect on the supplementary innovation inputs on the basis of the failure of firm innovation, which means the hypothesis 1a hasn’t been verified. As for the model (4), the F value is 39.401 with the significance level P < 0.05, and it passes the F test. The risk preference of TMT (taken as the independent variable) has a significant positive effect on the new turn of innovation on the basis of the failure of firm innovation. The standardized regression coefficient is 0.087 (P < 0.05), which indicates that the higher the risk preference of TMT, the greater the new round of innovation inputs are, which verifies the hypothesis 1b.

Table 3

Model (2) – (4) empirical test results

	Dependent Variable
	Model (2)	Model (3)	Model (4)
	reinnovation	incremental	novel
mrp	0.107^***	0.33	0.087^**
	(0.027)	(0.391)	(0.032)
size	0.915^***	0.923	0.931^***
	(0.044)	(0.633)	(0.051)
dlds	–1.132	20.757^*	–2.318^**
	(0.797)	(11.595)	(0.942)
growth	–0.158	0.334	–0.091
	(0.174)	(2.525)	(0.205)
roe	2.496^***	4.651	2.46^***
	(0.555)	(8.072)	(0.656)
state	–0.052	–2.697^*	0.14
	(0.101)	(1.466)	(0.119)
year	Control	Control	Control
_cons	–2.265^**	–20.489	–2.358
	(0.928)	(13.499)	(1.097)
F	55.201	1.403	39.401
R²	0.738	0.067	0.667

Note: ^*, ^** and ^*** indicate the significant level of 10%, 5% and 1% respectively, and the standard error SE in brackets.

Secondly, we test the moderating effect of top managers’ pay gap on the relationship between the risk preference of TMT and the re-innovation behavior of firm innovation failure (hypothesis 2, 2a, 2b). The models to be tested are as follows:

$\begin{matrix} {reinnovation}_{(i, t + 1)} = α + β_{1} * {mrp}_{(i, t)} + β_{2} * {gap}_{(i, t)} \\ + β_{3} * mrp * {gap}_{(i, t)} + β_{4} * {size}_{(i, t)} \\ + β_{5} * {dlds}_{(i, t)} + β_{6} * {growth}_{(i, t)} \\ + β_{7} * {roe}_{(i, t)} + β_{8} * {state}_{(i, t)} + μ_{(i, t)} \end{matrix}$ (5) $\begin{matrix} {incremental}_{(i, t + 1)} = α + β_{1} * {mrp}_{(i, t)} + β_{2} * {gap}_{(i, t)} \\ + β_{3} * mrp * {gap}_{(i, t)} + β_{4} * {size}_{(i, t)} \\ + β_{5} * {dlds}_{(i, t)} + β_{6} * {growth}_{(i, t)} \\ + β_{7} * {roe}_{(i, t)} + β_{8} * {state}_{(i, t)} + μ_{(i, t)} \end{matrix}$ (6) $\begin{matrix} {novel}_{(i, t + 1)} = α + β_{1} * {mrp}_{(i, t)} + β_{2} * {gap}_{(i, t)} \\ + β_{3} * mrp * {gap}_{(i, t)} + β_{4} * {size}_{(i, t)} \\ + β_{5} * {dlds}_{(i, t)} + β_{6} * {growth}_{(i, t)} \\ + β_{7} * {roe}_{(i, t)} + β_{8} * {state}_{(i, t)} + μ_{(i, t)} \end{matrix}$ (7)

Among them, reinnovation_(i,t+1), incre - linebreakmental_(i,t+1), novel_(i,t+1) represent the i-th sample firm in period t + 1’s re-innovation input after the failure of innovation, brand-new innovation, respectively, as well as the supplementary innovation. mrp_(i,t) indicates the risk preference of TMT of the i-th sample firm in t period. gap_(i,t) indicates the top managers’ pay gap of the i-th sample firm in t period. mrp * gap_(i,t) indicates the moderating effect of pay gap in the period t of the i-th sample firm. size_(i,t), dlds_(i,t), growth_(i,t), roe_(i,t), state_(i,t) represent the i-th sample firm in period t’s firm scale, proportion of independent directors, firm growth, nature of firm ownership respectively, as well as the returns on assets. β₁ to β₈ are regression coefficients, indicating the influence degree of independent variable, moderator variable and control variable on dependent variable. α is constant term, μ_(i,t) is random disturbance term.

According to the series of results shown in Table 4, the standardized regression coefficient of the cross-product term (mrp*gap) is 0.107 (P < 0.01) in model (5), and hypothesis 1 is assumed to be true, which indicates that pay gap between the top managers exerts positive moderating effect on the relationship between the risk preference of TMT and the re-innovation inputs after the failure of firm innovation, thus hypothesis 2 is verified. As for model (6), the cross-product term (mrp*gap) is 1.086 (P < 0.05), while hypothesis 1a is assumed to be false, which can only indicate that both pay gap between the top managers and the risk preference of TMT have positive effects on the supplementary innovation inputs without any moderating effect, so hypothesis 2a is not verified. As for model (7), the standardized regression coefficient of the cross-product term (mrp*gap) is 0.126 (P < 0.01), and hypothesis 1b is assumed to be true, which indicates that pay gap between the top managers has positive moderating effect on the relationship between the risk preference of TMT and the brand-new innovation after the failure of firm innovation, so hypothesis 2b is verified.

Table 4

Model (5) – (7) empirical test results

	Dependent variable
	Model (5)	Model (6)	Model (7)
	reinnovation	incremental	novel
mrp	–0.012	0.767	0.045
	(0.044)	(0.644)	(0.052)
gap	–0.82^***	–7.212^*	–0.734^**
	(0.256)	(3.79)	(0.303)
mrp^*gap	0.119^***	1.086^**	0.126^***
	(0.035)	(0.513)	(0.041)
size	0.922^***	0.938	0.918^***
	(0.044)	(0.656)	(0.052)
dlds	–1.117	20.907^*	–2.295^**
	(0.78)	(11.527)	(0.923)
growth	–0.117	0.755	–0.033
	(0.78)	(2.52)	(0.202)
roe	2.573^***	5.217	2.476^***
	(0.545	(8.056)	(0.645)
state	–0.052	–2.664^*	0.156
	(0.099)	(1.465)	(0.117)
year	Control	Control	Control
_cons	–1.634^***	–14.024	–1.366
	(0.949)	(14.021)	(1.122)
F	49.714	1.552	35.647
R²	0.751	0.086	0.684

Note: ^*, ^** and ^*** indicate the significant level of 10%, 5% and 1% respectively, and the standard error SE in brackets.

In terms of control variables, combined with models (2) – (7), we can see that firm scale and return on assets have positive correlation with re-innovation input and brand-new innovation after innovation failure, but not with supplementary innovation. However, the proportion of independent directors has positive correlation with supplementary innovation, while -is significantly negatively correlated with brand-new innovation.

4.2 Robustness test

In order to verify the robustness of the regression results, this paper tests relevant assumptions as follows:

(1) Change the estimation method, and use the generalized least square method of FGLS feasibility to re-test;

(2) Considering the omitted variables may affect the result of the estimated results, this paper employs the methods of Huang[34] for reference. The paper adopts the methods of adding multiple control variables and adding other variables that may affect firm re-innovation behavior on the basis of the original control variables, such as the dual role of the chairman and general manager (id), the separation of two rights (sr) and so on. What’s more, it also adopts the same regression method to re-evaluate to these models as to the original models without adding new control variables.

Through Tables 5 and 6, it is found that there is no substantial difference between the estimation results and the original basic regression estimation results, whether the model is replaced or the robustness test is conducted by adding multiple control variables or not. Therefore, it can be considered that the research conclusion is relatively stable.

Table 5
FGLS regression analysis

Dependent variable

Model (1) Model (2) Model (3) Model (4) Model (5) Model (6)

reinnovation incremental novel reinnovation incremental Novel

mrp 0.084^* 0.34 0.074^* –0.034 0.702 0.05

(0.021) (0.305) (0.025) (0.039) (0.569) (0.046)

gap –0.811^*** –7.047^* –0.73^**

(0.25) (3.671) (0.294)

mrp^gap 0.12^ 1.065^ 0.125^*

(0.034) (0.496) (0.04)

size 0.899^* 0.964 0.92^* 0.906^* 1.014^* 0.915^***

(0.041) (0.59) (0.048) (0.041) (0.602) (0.048)

dlds –0.974 20.191^* –2.204^** –0.99 20.031^* –2.237^

(0.772) (11.159) (0.907) (0.751) (11.047) (0.884)

growth –0.137 0.34 –0.074 –0.098 0.691 –0.029

(0.169) (2.439) (0.198) (0.164) (2.419) (0.194)

roe 2.637^* 4.307 2.56^* 2.678^* 4.582 2.523^***

(0.533) (7.705) (0.626) (0.522) (7.678) (0.614)

state –0.068 –2.684^* 0.132 –0.064 –2.629^* 0.153

(0.098) (1.422) (0.116) (0.096) (1.418) (0.113)

_cons –1.971^** –21.373^* –2.176^** –1.361 –15.732 –1.321

(0.879) (12.706) (1.033) (0.883) (12.993) (1.04)

Chi-square 628.344 15.436 455.167 676.767 20.469 491.754

	Dependent variable
mrp	0.084^***	0.34	0.074^***	–0.034	0.702	0.05
	(0.021)	(0.305)	(0.025)	(0.039)	(0.569)	(0.046)
gap				–0.811^***	–7.047^*	–0.73^**
				(0.25)	(3.671)	(0.294)
mrp^*gap				0.12^***	1.065^**	0.125^***
				(0.034)	(0.496)	(0.04)
size	0.899^***	0.964	0.92^***	0.906^***	1.014^*	0.915^***
	(0.041)	(0.59)	(0.048)	(0.041)	(0.602)	(0.048)
dlds	–0.974	20.191^*	–2.204^**	–0.99	20.031^*	–2.237^**
	(0.772)	(11.159)	(0.907)	(0.751)	(11.047)	(0.884)
growth	–0.137	0.34	–0.074	–0.098	0.691	–0.029
	(0.169)	(2.439)	(0.198)	(0.164)	(2.419)	(0.194)
roe	2.637^***	4.307	2.56^***	2.678^***	4.582	2.523^***
	(0.533)	(7.705)	(0.626)	(0.522)	(7.678)	(0.614)
state	–0.068	–2.684^*	0.132	–0.064	–2.629^*	0.153
	(0.098)	(1.422)	(0.116)	(0.096)	(1.418)	(0.113)
_cons	–1.971^**	–21.373^*	–2.176^**	–1.361	–15.732	–1.321
	(0.879)	(12.706)	(1.033)	(0.883)	(12.993)	(1.04)
Chi-square	628.344	15.436	455.167	676.767	20.469	491.754

Note: ^*, ^** and ^*** indicate the significant level of 10%, 5% and 1% respectively, and the standard error SE in brackets.

Table 6

Regression analysis after adding new control variables

	Dependent variable
	Model (1)	Model (2)	Model (3)	Model (4)	Model (5)	Model (6)
	reinnovation	incremental	Novel	reinnovation	incremental	Novel
mrp	0.109^***	0.379	0.087^***	–0.016	0.781	0.056
	(0.027)	(0.386)	(0.032)	(0.043)	(0.636)	(0.051)
gap				–0.876^***	–8.197^**	–0.747^**
				(0.253)	(3.74)	(0.302)
mrp^*gap				0.126^***	1.171^**	0.134^***
				(0.034)	(0.505)	(0.041)
size	0.922^***	1.11^*	0.925^***	0.928^***	1.182^*	0.903^***
	(0.044)	(0.634)	(0.052)	(0.045)	(0.665)	(0.054)
dlds	–1.823^**	9.736	–2.716^***	–0.186^**	9.37	–2.716^***
	(0.828)	(12.01)	(0.987)	(0.807)	(11.925)	(0.962)
growth	–0.112	1.16	–0.071	–0.064	1.601	–0.011
	(0.173)	(2.503)	(0.206)	(0.169)	(2.492)	(0.201)
roe	2.32^***	2.151	2.322^***	2.388^***	2.809	2.303^***
	(0.552)	(8.004)	(0.658)	(0.54)	(7.972)	(0.643)
state	–0.072	–3.077^**	0.135	–0.073	–3.101^**	0.159
	(0.1)	(1.45)	(0.119)	(0.098)	(1.451)	(0.117)
id	0.173^*	3.423^**	0.035	0.178^*	3.485^**	0.001
	(0.091)	(1.317)	(0.108)	(0.091)	(1.339)	(0.108)
sr	1.069^*	10.939	1.281^*	1.222^**	12.305	1.641^**
	(0.56)	(8.116)	(0.667)	(0.555)	(8.199)	(0.662)
year	Control	Control	Control	Control	Control	Control
_cons	–2.913^**	–21.532	–2.086	–1.506	–15.313	–0.827
	(0.932)	(13.517)	(1.111)	(0.962)	(14.212)	(1.147)
F	48.48	1.902	33.918	45.186	2.032	31.91
R²	0.747	0.104	0.673	0.762	0.126	0.693

Note: ^*, ^** and ^*** indicate the significant level of 10%, 5% and 1% respectively, and the standard error SE in brackets.

5 Conclusions and implications

5.1 Conclusions

In view of the current high failure rate of innovation of Chinese firms, this paper takes the listed pharmaceutical manufacturing firms in China’s A-share market in 2011–2017 as the research sample. From the perspective of top managers’ pay gap, empirically tests the effect of risk preference of TMT on the re-innovation behavior after innovation failure, and finally draws the following conclusions:

(1) The risk preference of TMT is positively related to the re-innovation input and brand-new innovation after the innovation failure, but not to the supplementary innovation. That is to say, when the innovation failure occurs, the higher the risk preference of TMT, the greater the re-innovation input and brand-new innovation input after the failure, but the impact on the supplementary innovation is not significant. (2) The pay gap has positive moderating effect on the positive correlation between the risk preference of TMT and the re-innovation input and brand-new innovation after the failure of innovation. That is to say, the larger the pay gap is, the stronger the positive impact of the risk preference of TMT on the re-innovation input and brand-new innovation input after the failure of innovation.

The contribution of this paper is mainly reflected in the following aspects. Firstly, It expands the research on TMT and innovation failure. Secondly, From the perspective of executive compensation gap, the research on the relationship between the risk preference of TMT and the re-innovation behavior after firm innovation failure is helpful to improve the re-innovation behavior of TMT after innovation failure and improve the success rate of re-innovation after innovation failure by setting a reasonable compensation gap.

From the perspective of pay gap, this paper studies the impact of risk preference of TMT on re-innovation behavior after innovation failure, which to some extent makes up for the shortage of lacking for empirical research on innovation failure. From the perspective of firms, the research results put forward improvement on re-innovation behavior after innovation failure, which has certain theoretical and practical significance. However, there are still some limitations in this study: (1) Limitations of research samples. In the selection of empirical research samples, only the relevant data of Listed Companies in the pharmaceutical manufacturing industry is selected as the research sample of this paper, which fails to realize the goal of collection and research of the whole industry research sample data. Therefore, the research conclusions of this paper may reflect more of the impact of risk preference of TMT on the re-innovation behavior after the failure of innovation in the pharmaceutical manufacturing industry, and fail to reflect the characteristics of the whole industry in this respect, which leads to certain impact and limitations on the representativeness and universality of the research results. (2) Limitations of research variables. In view of China’s unique economic system, institutional environment such as government intervention, financial development level, it can be speculated that legal environment and cultural environment may also have impact on the technological innovation activities of firms, as well as the willingness and behavior of re-innovation after failure. Therefore, more comprehensive consideration should be given to the impact of external environment on subsequent innovation behavior. In the future research, in order to further explore the characteristics of innovation failure enterprises in the whole industry and find more appropriate group decision making model methods research innovation failure or more appropriate variables to measure innovation failure. By seeking the help of government departments, enterprises are more willing to cooperate with or disclose the relevant data of innovation failure, and build the innovation failure database. It lays the data foundation for the follow-up research, and provides strong theoretical support and decision-making reference for the re innovation development of enterprises with technological innovation failure.

5.2 Implications

According to the conclusion of the previous study, with the increase of risk preference of TMT, the re-innovation input and brand-new innovation will increase after the failure of innovation. Because the sum of the brand-new innovation and the supplementary innovation is the re-innovation input, it shows that when the innovation fails, the TMT is more inclined to carry out the brand-new innovation with higher risk coefficient. And with the increase of executive compensation gap, the probability of brand-new innovation will also increase. In view of the huge sunk cost of the firm’s early investment, and the supplementary innovation is the supplementary investment to the original innovation failure project based on the existing human, material and technical resources, and the brand-new innovation often has greater uncertainty and risk coefficient than the supplementary innovation, and needs more human, financial and material resources. And most of the firms’ innovation failure only represents the failure of a certain stage of the firm, and does not mean the bankruptcy and death of the firm. A project that fails in technological innovation often contains great potential value, so it can not be ignored or treated negatively because of its “failure” attribute. Therefore, the TMT should not blindly carry out brand-new innovation activities, but should pay more attention to the failed projects and tap their potential value. In order to reduce the over investment and waste of resources and improve the re-allocation rate of resources, firms can restrain the brand-new innovation behavior after the failure of innovation by narrowing the salary gap among the members of the TMT, and use the limited resources of the enterprise as the supplementary investment of the failed innovation projects, so as to reduce the excessive investment and waste of resources, and improve the re-allocation rate of resources, so as to improve the success rate of re-innovation.

Footnotes

Acknowledgment

This study was supported by the National Social Science Fund of China (Grant No. 19BGL039).

References

Jordan

A.H.

and Audia

P.G.

, Self-enhancement and learning from performance feedback, Academy of Management Review 37(2) (2012), 211–231.

Abdel-Khalik

A.R.

, Prospect theory predictions in the field: risk seekers in settings of weak accounting controls, Journal of Accounting Literature 33(1-2) (2014), 58–84.

Xie

, The impact of TMT characteristics on performance of listed companies on GEM, Master’s thesis, Harbin University of Science and Technology, 2016.

Mandl

, Berger

C.S.E.

and Kuckertz

, Do you plead guilty? Exploring entrepreneurs’ sense making-behavior link after business failure, Journal of business venturing Insights (5) (2016), 9–13.

Maslach

, Change and persistence with failed technological innovation, Strategic Management Journal 37 (2016), 714–723.

Ucbasaran

, Lockett

, Wright

, et al., Entrepreneurial founder teams: factors associated with member entry and exit, Entrepreneurship Theory & Practice 28(2) (2010), 107–128.

H.M.

, TMT’s characteristics and investment efficiency: moderating effect of TMT incentives mechanism, Jinan University, 2016.

H.Z.

and Liu

, Market characteristics of high technology and its products from the perspective of technology life cycle, Science Research Management 3 (1993), 33–35.

Perez

I.J.

, Cabrerizo

F.J.

and Herrera-Viedma

E.A.

, A mobile group decision making model for heterogeneous information and changeable decision contexts, International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems 19(supp01) (2011), 33–52.

10.

Morente-Molinera

J.A.

, Rios-Aguilar

and Gonzalez-Crespo

, Dealing with group decision making environments that have a high amount of alternatives using card-sorting techniques, Expert Systems with Applications 127(8) (2019), 187–198.

11.

Denrell

, Organizational risk taking: adaptation versus variable risk preferences, Industrial and Corporate Change 17(3) (2008), 427–466.

12.

and Song

Z.G.

, The key driving factors of tech sme technology innovation; An exploratory analysis based on 4 Tech SMES in Beijing and Tianjin, Science of Science and Management of S.&T 35(05) (2014), 156–163.

13.

Zhang

and Duan

Y.L.

, An empirical study on the impact of market-oriented equilibrium on product innovation performance of manufacturing firms, Management World (12) (2010), 119–130.

14.

Cope

, Entrepreneurial learning from failure: An interpretative phenomenological analysis, Journal of Business Venturing 26(6) (2011), 604–623.

15.

Eggers

J.P.

, Falling flat: failed technologies and investment under uncertainty, Administrative Science Quarterly 57 (2012), 47–80.

16.

Zhu

J.W.

, Peng

J.J.

and Liu

, Study on impact of characteristics of top management team on firm technology innovation investment: moderating effect of team incentives, Scientific Decision (08) (2014), 17–33.

17.

Zhu

L.J.

, Cash dividend, executive gap and firm performance-Based on the A-share of manufacturing data, Southwest University, 2016.

18.

Cain

M.D.

and McKeon

S.B.

, CEO personal risk-taking and corporate policies, SSRN Electronic Journal, 2013.

19.

, A study on the influence of risk preference of senior management team on firm innovation performance-based on the empirical data of listed companies on GEM, Master’s thesis, Hebei Agricultural University, 2018.

20.

Xing

and Wang

G.H.

, A study on the innovation path of technology-based incubated firm from the perspective of entrepreneurial orientation, Science of Science and Management of S. & T 7 (2014), 116–127.

21.

Cyert

R.M.

, March

J.G.

, A Behavioral Theory of the Firm, Blackwell Publishers Inc.: Malden, MA, 1963.

22.

Wang

W.H.

, Zhang

and Ji

X.L.

, Managerial owership and R&D investment: Interest convergence effects or managerial entrenchment effects—An empirical study based on the listed high-tech companies, R&D Management 26(04) (2014), 23–31.

23.

Zhang

W.T.

, Financial risk control of small and medium-sized firms, Modern Economic Information (22) (2013), 279+282.

24.

and Su

X.H.

, Study on background characteristics of top management team, incentive mechanism and high-tech firm’s R&D investment—Evidence from listed high-tech firms of A-shares, Science and Technology Management Research 32(06) (2012), 100–108.

25.

X.Y.

, Pu

X.L.

and Li

, Internal cause? External cause? Don’t be so quick to assign blame for a startup’s failure, Tsinghua Business Review (5) (2017), 48–54.

26.

Belghitar

and Clark

E.A.

, The effect of CEO risk appetite on firm volatility: an empirical analysis of financial firms, Routledge (2012), 195–211.

27.

Bai

Y.T.

, Guo

J.E.

and Xi

Y.M.

, The effects of top management team’s risk preference diversity on strategic investment decision making: an experiment research, Nankai Business Review 10(12) (2007), 25–30+44.

28.

Y.X.

, Fu

, Li

P.F.

and Feng

B.J.

, Literature survey on risk preference experiment, Science Technology and Management 11(05) (2009), 34–37.

29.

Xiong

, Ye

J.M.

and Wang

P.J.

, Does the institutional environment affect the failed technological innovation in firms? Evidence from listed companies in China’s pharmaceutical manufacturing industry, Transformations in Business & Econmics 18(1) (2019), 60–80.

30.

Wen

Z.L.

, Hou

J.T.

and Zhang

, A comparison of moderator and mediator and their applications, Acta Psychologica Sinica 37(2) (2005), 268–274.

31.

Xiong

, Wang

P.J.

and Zhao

, Re-innovation from failure, institutional environmental differences, and firm performance: evidence from China, Amfiteatru Economic 22(53) (2020), 197–219.

32.

Xiong

, Intelligent simulation of enterprise re-innovation support system based on system dynamics, Journal of Intelligent and Fuzzy Systems 38(6) (2020), 1–11.

33.

Gong

Z.W.

, Xu

X.X.

and Guo

W.W.

, Minimum cost consensus modelling under various linear uncertain-constrained scenarios, Information Fusion 66 (2021), 1–17.

34.

Huang

H.G.

, Bai

C.Q.

and Yi

, Financial Exclusion, investment in agricultural science and technology and agricultural economic development, Management World 34(09) (2018), 67–78.