INTRODUCTION
Corporate innovation is an essential driver of a firm's competitive advantage and an engine of economic development that also improves our way of living and contributes to societal advancement (Ahuja, Lampert, & Tandon, Reference Ahuja, Lampert and Tandon2008; Schumpeter, Reference Schumpeter1911). Among voluminous inquiries seeking to understand the determinants of corporate innovation, a stream of studies examine how performance feedback affects corporate innovation strategy. Growing out of Cyert and March (Reference Cyert and March1963)'s behavioral theory of the firm, performance feedback theory posits that firm decisions are guided by aspirations. When performance falls short of aspirations, firms are triggered to conduct problemistic search to find solutions to remedy the underperforming problem. In contrast, when performance is above aspiration levels, firms tend to maintain the status quo and follow previous routines (Gavetti, Greve, & Levinthal, Reference Gavetti, Greve, Levinthal and Ocasio2012). Although it is predicted that negative performance feedback prompts organizational change, firms differ in how they interpret and react to this information. Empirically, organizational scholars observe that some firms attempt to close the performance gap by increasing investment in innovation (Chen & Miller, Reference Chen and Miller2007; Greve, Reference Greve2003), while others choose not to do so but instead implement cost-cutting, resource conservation, or other performance improvement strategies (Bromiley & Washburn, Reference Bromiley and Washburn2011; Greve, Reference Greve2010; Shimizu, Reference Shimizu2007).
A natural question then is why organizations adopt different innovation strategies in response to negative feedback and what affects their choices. Extant literature has articulated that the selection of aspiration levels (Lucas, Knoben, & Meeus, Reference Lucas, Knoben and Meeus2018; Lv, Chen, Zhu, & Lan, Reference Lv, Chen, Zhu and Lan2019), magnitude of performance discrepancies (Hu, Blettner, & Bettis, Reference Hu, Blettner and Bettis2011; Shimizu, Reference Shimizu2007), and duration of underperformance (Yu, Minniti, & Nason, Reference Yu, Minniti and Nason2019) all affect organizational responses. Heterogeneity in organizational characteristics, such as the abundance of organizational slack (Chen & Miller, Reference Chen and Miller2007; Kuusela, Keli, & Maula, Reference Kuusela, Keli and Maula2017), firm size (Audia & Greve, Reference Audia and Greve2006; Greve, Reference Greve2010), firm age (Blettner, He, Hu, & Bettis, Reference Blettner, He, Hu and Bettis2015), firm capability (Eggers & Kaul, Reference Eggers and Kaul2018), technological experience (Eggers & Suh, Reference Eggers and Suh2019), board structure (Desai, Reference Desai2016), and organizational structure (Gaba & Joseph, Reference Gaba and Joseph2013; Joseph, Klingebiel, & Wilson, Reference Joseph, Klingebiel and Wilson2016) also contribute to distinct firm responses.
We extend this research stream by considering managerial cognition as a key factor shaping firms’ responses to negative performance feedback. Drawing upon insights from the attention-based view (Ocasio, Reference Ocasio1997), we argue that whether firms will increase research and development (R&D) investment in response to performance shortfalls is affected by decision-makers’ attention allocation. Decision-makers only act on issues and utilize solutions upon which they focus their attention, and their attention allocation is channeled by their subjective interpretation of situations (Dutton & Jackson, Reference Dutton and Jackson1987; Ocasio, Rhee, & Milner, Reference Ocasio, Rhee, Milner, Argote and Levine2020). That is, managerial cognition affects managers’ interpretation of performance feedback, shapes their selective attention, influences their perception of innovation as a valid solution to close performance gaps, and consequently guides firms’ strategic responses to performance shortfalls. We then explore how managerial cognition, as shaped by managers’ experiences, connections, positions, and industry environments, affects firms’ allocation of attention and, consequently, their decisions to invest in innovation.
Prior empirical studies that built on performance feedback theory have mainly focused on how variations in aspiration levels affect firms’ responses to performance feedback. In contrast, our article highlights the essential role of cognition and attention allocation in channeling firms’ action choices. We argue that organizational search is not only influenced by aspirations generated from a backward-looking, feedback-based process but also involves a forward-looking, cognitive process. Managerial decisions are based on observations and interpretations of the environment that direct decision-makers’ attention to a particular solution, such as innovation (Gavetti & Levinthal, Reference Gavetti and Levinthal2000). According to Ocasio (Reference Ocasio1997: 189), ‘selective attention’ refers to a decision-maker's focus of time and effort on both issues that represent an ‘available repertoire of categories for making sense of the environment’ and answers from an ‘available repertoire of actions alternatives’. Because decision-makers are inherently limited in their attentional capacity, they discern the relative importance of issues and answers, select certain issues and answers to focus their attention on, and take action accordingly (Ocasio et al., Reference Ocasio, Rhee, Milner, Argote and Levine2020).
On the one hand, negative performance feedback directs decision-makers’ attention to performance problems and raises the urgency to address those problems. On the other hand, decision-makers attend to different solutions to remedy poor performance. Our perspective is that managers’ subjective evaluation of innovation as a valid performance-enhancing solution is dependent on dispositional and situational factors that steer their allocation of attention to innovation. As such, our study addresses contradictory empirical findings related to the impact of performance shortfalls on firm innovation from a cognitive lens. We explore how internal and external contingencies shape managerial cognition and affect firms’ attention allocation and courses of action.
Our study enriches the notion of ‘situated attention’ (Ocasio, Reference Ocasio1997) in the attention-based view by highlighting managerial cognition as the micro-foundation of organizational attention. The social cognitive theory posits that managerial cognition is constrained by the environment (Moskowitz, Reference Moskowitz2005). Managers are embedded in a social context; thus, the interaction between managers and their environment plays a significant role in shaping their cognitive process, guiding their allocation of attention, and subsequently influencing the trajectory of organizational change. Using managerial cognition as a pillar to explain performance feedback-based learning enables the consideration of both rational and irrational elements in managerial decision-making. Through this cognitive lens, our article unveils how organizations learn from the past, evaluate future opportunities, and choose courses of present action. Our study thus also advances managerial cognition research to highlight the essential role of top managers’ cognition in directing firm behavior (e.g., Eggers & Kaplan, Reference Eggers and Kaplan2009; Kaplan, Reference Kaplan2008).
The remainder of the article is organized as follows. The section ‘Theoretical foundation and hypotheses development introduces the theoretical foundation of the article and builds the hypotheses. In the section ‘Methods’, we describe the data, variables, and empirical models. We present our results in the section ‘Results’ and conclude the article with a discussion in the section ‘Discussion’.
THEORETICAL FOUNDATION AND HYPOTHESES DEVELOPMENT
The behavioral theory of the firm acknowledges that decision-makers only possess bounded rationality and utilize a ‘satisficing’ instead of a maximizing decision-making rule (Cyert & March, Reference Cyert and March1963; March & Simon, Reference March and Simon1958). What is deemed as ‘satisfactory’ is dependent on organizational aspirations set by comparing the organization's current performance with those of socially comparable peers and the firm's own historical performance (Gavetti et al., Reference Gavetti, Greve, Levinthal and Ocasio2012). When firm performance falls below aspiration levels, problemistic search is triggered for firms to find solutions to address performance shortfalls by changing strategies, routines, and practices (Posen, Kell, Kim, & Meissner, Reference Posen, Kell, Kim and Meissner2018). Such searches and changes are less likely to happen when performance is above the aspiration level. Performance feedback-based organizational learning is inherently about attention allocation, through which organizational members filter out some issues and only dedicate cognitive resources to a few selected problems and solutions (Ocasio et al., Reference Ocasio, Rhee, Milner, Argote and Levine2020). Building on this premise, Ocasio (Reference Ocasio1997: 189) developed the attention-based view, where he defined attention as a cognitive process encompassing ‘noticing, encoding, interpreting, and focusing of time and effort’. The attention-based view is centered around three principles: (1) focus of attention: indicating that decision-makers act on issues and answers upon which they focus their attention; (2) structural distribution of attention: highlighting the roles of organizational rules, resources, and structures in channeling the focus of decision-makers’ attention; and (3) situated attention: emphasizing the influence of contextual factors on attention allocation (Ocasio, Reference Ocasio2011).
When performance falls below the aspiration level, decision-makers will switch their attention to the performance issue and engage in problemistic search to improve performance to reach the target level (Cyert & March, Reference Cyert and March1963). Corporate innovation is a vital factor for firms to differentiate from competitors. Thus, underperforming firms may view R&D investment as an effective means to revitalize their businesses and to improve performance by upgrading technology and product portfolio. Consequently, these firms may respond to negative performance feedback by increasing resource allocation to R&D, hoping to generate a fair return on their investment (Chen & Miller, Reference Chen and Miller2007; Greve, Reference Greve2003). Underperformance might also prompt executives to form more optimistic expectations of gains from risky innovation projects, thereby motivating them to engage in risky innovation investment (Greve, Reference Greve1998; Greve & Taylor, Reference Greve and Taylor2000).[Footnote 1] We thus make the following baseline prediction:
Hypothesis 0: Performance shortfalls will lead to increases in R&D investment.
Top management plays a significant role in formulating firm strategy and shaping changes (Hambrick & Finkelstein, Reference Hambrick and Finkelstein1987). Managerial cognition literature has long recognized that strategic action is shaped by how managers notice and interpret informational cues in the environment, assess opportunities, and translate their subjective perspectives into courses of action (e.g., Cho & Hambrick, Reference Cho and Hambrick2006; Kaplan, Reference Kaplan2008). Thus, what top managers pay attention to shapes how they decide and act (Ocasio, Reference Ocasio1997). Because executives only possess limited information-processing capability, they are unable to evaluate all solutions when conducting problemistic search and must resort to a specific set of sensory solutions for answers (March & Simon, Reference March and Simon1958).
The social cognitive theory argues that decision-makers utilize their previously developed knowledge and accumulated experiences to process most information, form expectations, and make decisions (Moskowitz, Reference Moskowitz2005). Specifically, executives’ functional background creates a lens through which they perceive and interpret business problems and assess the viability of different strategic options (Hambrick & Mason, Reference Hambrick and Mason1984). For example, CEOs with significant career experience in output functions such as R&D, engineering, marketing, and sales are found to favor an innovation strategy because they prefer to achieve growth through discovering new products and markets, while executives with extensive experience in throughput functions such as accounting, finance, and administration are more interested in process improvement (Finkelstein, Hambrick, & Cannella, Reference Finkelstein, Hambrick and Cannella2009). Among CEOs with experience in output functions, those with R&D and engineering backgrounds are found to be more keen on investing in R&D than those with marketing and sales backgrounds (Barker & Mueller, Reference Barker and Mueller2002).
Although innovation presents underperforming firms with an opportunity to close the performance gap, this opportunity is generally fraught with uncertainty. The uncertainty problem is further intensified when negative performance feedback demands immediate firm attention and action. The managerial cognition literature suggests that executives experiencing uncertainty and tight decision deadlines often employ cognitive shortcuts, or heuristics, to make reasonable inferences to solve complex and uncertain problems (Moskowitz, Reference Moskowitz2005). When facing urgent but ambiguous situations, executives tend to resort to familiar domains to seek solutions, and their familiarity with the current situation influences their preferences for alternative courses of action, which is called the ‘familiarity effect’ (Heath & Tversky, Reference Heath and Tversky1991). For example, Eggers and Song (Reference Eggers and Song2015) find that serial entrepreneurs experiencing a venture failure tend to rely on existing strategies they are familiar with and are less likely to alter their managerial styles and strategies. Eggers and Suh (Reference Eggers and Suh2019) likewise notice that when receiving negative performance feedback from technical domains in which they have experience, executives are prone to increase efforts and investments in these familiar domains in the hope of a turnaround.
Day and Lord (Reference Day and Lord1992) argue that individuals with significant experience are more schema-driven than novices, and their decision outcomes are heavily influenced by heuristics developed through past experiences. We thus expect that CEOs with an R&D and engineering background are likely to resort to their functional expertise to make sense of and seek solutions to address performance shortfalls. As a result, these executives may interpret underperformance as a lack of successful innovation and decide to allocate more resources to innovation. Moreover, CEOs with technical backgrounds may form more positive expectations on returns on innovation than executives without such backgrounds. Owing to their familiarity with product and process innovation, CEOs with R&D or engineering backgrounds are thus more likely to focus their attention on innovation as a valid performance-enhancing solution and choose to increase R&D investment to remedy poor performance. Taken together, we make the following prediction:
Hypothesis 1: Firms whose CEOs have R&D or engineering backgrounds are associated with a larger R&D investment increase in response to performance shortfalls than firms with CEOs of other functional backgrounds.
The assumptions underlying problemistic search suggest that managerial experience and the information available to managers for decision-making affect firms’ search trajectories and influence how resources are orchestrated and deployed to solve problems (Eggers & Kaplan, Reference Eggers and Kaplan2013). That is, managers’ track records shape how they interpret causes, draw inferences, make attributions, and form expectations, which we argue guide their attention toward a particular solution to address performance problems. When receiving negative performance feedback, executives tend to find solutions in familiar domains. For example, Shimizu (Reference Shimizu2007) finds that firms with divestiture experience are more inclined to implement this strategy when facing performance shortfalls. Gaba and Joseph (Reference Gaba and Joseph2013) document that business unit managers facing performance downturns seek solutions in familiar product markets by rationalizing manufacturing, improving support systems, and increasing new product offerings. In contrast, corporate managers tend to apply broad cost-cutting solutions derived from their portfolio management experience by reducing discretionary spending and decreasing labor costs.
In a transitional economy such as China, the government controls a significant proportion of strategic factor resources and possesses considerable power to allocate financial resources and provide favored treatments to selected organizations (Conyon, He, & Zhou, Reference Conyon, He and Zhou2015; Li, Meng, Wang, & Zhou, Reference Li, Meng, Wang and Zhou2008). The government's preferential treatment is geared toward firms whose executives possess political ties (Peng & Luo, Reference Peng and Luo2000). Politically connected firms often enjoy priority access to equity markets and bank financing, government subsidies and tax benefits, lavish public contracts, land permits, export licenses, and other scarce resources controlled by the government (Cull, Li, Sun, & Xu, Reference Cull, Li, Sun and Xu2015; Faccio, McConnell, & Masulis, Reference Faccio, McConnell and Masulis2006). Returns from building and strengthening political connections thus can be highly significant for these firms (Li & Zhang, Reference Li and Zhang2007).
When receiving negative performance feedback, politically connected CEOs who have traditionally benefited from preferential treatment from the government may again resort to the government to solve their performance problems by seeking government assistance such as bailouts, financial support, subsidies, new contracts, and supporting policies. Because executives in politically connected firms are more inclined to rely on political patronages for assistance, they are less interested in more time-consuming and risky investments such as innovation projects (Wu, Reference Wu2011). Therefore, we predict that politically connected CEOs may view strengthening government relationships as a preferable performance-enhancing solution to close the performance gap rather than investing in innovation. Moreover, innovation is not the most familiar domain for these executives and thus is unlikely to receive managerial attention in times of uncertainty and urgency. Finally, politically connected firms may be asked to satisfy the government's noneconomic goals, such as maintaining the employment level in a business downturn, in return for special treatment that they receive from the government (Fan, Wong, & Zhang, Reference Fan, Wong and Zhang2007). These additional constraints further prohibit politically connected firms from allocating resources to long-term innovation projects in the event of poor performance. We thus make the following prediction:
Hypothesis 2: Firms whose CEOs are politically connected are associated with a smaller R&D investment increase in response to performance shortfalls than firms without politically connected CEOs.
Executives’ cognition and attention are also affected by their positions within their organization. Ocasio (Reference Ocasio1997) argues that an organization can be perceived as a system of procedures and communication channels that receives and processes information from its environment, feeding this information to decision-makers and retrieving alternative solutions from the environment. An organizational context that decision-makers find themselves in, thus, shapes their focus of attention, defines the trajectory of problemistic search, and guides and constrains how organizational tasks are accomplished (Gavetti, Greve, Levinthal, & Ocasio, Reference Gavetti, Greve, Levinthal and Ocasio2012; Ocasio, Reference Ocasio1997).
In modern corporations with a separation of ownership and control, CEOs make key strategic decisions under the active monitoring of boards of directors who also provide advice to top management (Adams & Ferreira, Reference Adams and Ferreira2007). Therefore, corporate decisions are affected by the board structure that defines the relationship between the CEO and the board of directors (Desai, Reference Desai2016). A separate leadership structure with different individuals holding the CEO and the board chair position distributes power between top commanders of the firm using a check-and-balance design. In contrast, a combined structure with the same person holding both leadership positions is characterized by the concentrated power and unity of command (Finkelstein & D'Aveni, Reference Finkelstein and D'Aveni1994). When a firm adopts a separate leadership structure, the CEO and the board chair may hold different opinions regarding an appropriate response to performance shortfalls (Dowell, Shackell, & Stuart, Reference Dowell, Shackell and Stuart2011). Although the chairperson can offer additional advice and recommend alternative solutions to the CEO, the division of power between these individuals makes consensus more difficult and decision-making more time-consuming and risk-averse. Contradictory perspectives provided by the board chair and CEO may also complicate a firm's formulation of any response strategies, particularly uncertain and risky ones, such as investing in innovation (Desai, Reference Desai2016).
In the event of performance shortfalls, managers need substantial discretion to engage in problemistic search and craft strategies, and deploy firm resources to address the underperformance problem. Extensive monitoring under a separate leadership structure reduces CEO discretion and diminishes their incentives to engage in explorative search and risky behaviors such as innovation investment (Krause, Semadeni, & Cannella, Reference Krause, Semadeni and Cannella2014). In contrast, boards with a combined leadership position devote less attention to monitoring and more to strategic advising, resource provisions, and other issues (Tuggle, Sirmon, Reutzel, & Bierman, Reference Tuggle, Sirmon, Reutzel and Bierman2010). He and Wang (Reference He and Wang2009) argue that a unified leadership structure enables CEOs to better utilize and deploy firm resources and increases their innovation management capacity. Hence, firms with this design benefit more from innovation. We thereby expect that a CEO who is also the board chair faces fewer internal conflicts, possesses larger discretion to formulate an innovation strategy, and has more power to deploy resources to implement this strategy than CEOs in firms with separate leadership positions. All these factors raise CEOs’ expectations on return on innovation investment and prompt them to remedy performance problems by increasing innovation investment.
In addition, a combined leadership structure may strengthen a CEO's self-confidence and pride (Li & Tang, Reference Li and Tang2010). Overconfident CEOs tend to underestimate ways in which risky strategies may fail, overestimate their problem-solving capabilities, exaggerate the potential benefits of a risky strategy, and are found to invest more in risky innovation projects (Galasso & Simcoe, Reference Galasso and Simcoe2011). The tendency to increase commitments to highly uncertain projects is also stronger when external oversight is limited, which is true in firms with a combined leadership structure (Vissa, Greve, & Chen, Reference Vissa, Greve and Chen2010). Therefore, we expect that CEOs who are also board chairs are more likely to increase innovation investment in case of poor performance compared with a less confident CEO whose power is constrained by a separate board chair.[Footnote 2] Taken together, we make the following prediction:
Hypothesis 3: Firms with a combined leadership structure are associated with a larger R&D investment increase in response to performance shortfalls than firms with a separated leadership structure.
Rooted in social cognitive theory, the principle of situated attention indicates that what decision-makers focus on and what they do depend on the social context in which they reside (Ocasio, Reference Ocasio1997). Industry is a powerful cognitive category, and industry boundaries play a salient role in influencing managerial cognition and shaping firm behaviors (Nadkarni & Barr, Reference Nadkarni and Barr2008). Industry structure delineates the relationship between the focal firm and other firms in the industry and affects how firms receive, perceive, and process information. As cognitive limits preclude top managers from forming a complete and accurate understanding of all strategic alternatives to address underperformance problems, executives may rely on their observations and interpretations of industrial environment to evaluate whether innovation is a valid performance-enhancing solution and decide how to react to negative performance feedback (Gavetti & Levinthal, Reference Gavetti and Levinthal2000).
In a concentrated industry dominated by a few large players with rich resources and strong market power, changes occur less frequently, typically along more predictable linear paths (Bogner & Barr, Reference Bogner and Barr2000). These industries are also simpler with more mature routines and norms (Li & Tang, Reference Li and Tang2010). As a result, executives in these industries pay less attention to innovation but instead build isolating mechanisms to deter imitations and protect existing core competencies (Garg, Walters, & Priem, Reference Garg, Walters and Priem2003). As continuous innovation is not a dominant strategy deployed by firms in these industries, underperforming firms in such industries are less likely to engage in extensive explorative search that entails a significant increase in innovation investment. In contrast, a competitive industry characterized by rapid and frequent environmental change requires top executives to actively conduct problemistic search to construct their environments through experimentation and innovation (Eisenhardt, Reference Eisenhardt1989). When encountering performance shortfalls, executives in these industries are likely to focus more attention on innovation and undertake more innovative activities to explore new opportunities to fix the performance problem compared with their counterparts in less competitive sectors (Nadkarni & Chen, Reference Nadkarni and Chen2014). We therefore make the following prediction:
Hypothesis 4: Firms competing in competitive industries are associated with a larger R&D investment increase in response to performance shortfalls than firms in concentrated industries.
The influence of industry on firm attention and decisions is also manifested through industry norms. Institutional theory suggests that firms may adopt strategies and practices of other organizations in the same industry, which is known as ‘mimetic isomorphism’ (DiMaggio & Powell, Reference DiMaggio and Powell1983). When uncertainty is high, social norms become an even more significant cognitive cue for decision-makers to draw inferences. Thus, imitating peers’ strategies and practices is particularly attractive in ambiguous and uncertain situations (Kostova & Roth, Reference Kostova and Roth2002). Cyert and March (Reference Cyert and March1963) maintain that such an imitation helps firms reduce search costs and decrease uncertainty.
Innovation investment is risky by nature in terms of the timing and degree of technological and market success (Souder & Shaver, Reference Souder and Shaver2010). Organizations are unsure about the probability of payoffs, time, and amounts when investing in innovation (Duran, Kammerlander, van Essen, & Zellweger, Reference Duran, Kammerlander, van Essen and Zellweger2016). When managers are unsure about the connections between their actions and outcomes, they are most likely to be receptive to information implicit in others’ actions (Lieberman & Asaba, Reference Lieberman and Asaba2006). As a result, managers may follow cues obtained from their industry peers by turning their attention to dominant industry practices and competitors’ strategies to cope with uncertainty (Hsieh, Tsai, & Chen, Reference Hsieh, Tsai and Chen2015). Performance below aspiration levels may also shake executives’ confidence in the effectiveness of their firms’ current strategy and practice. Therefore, executives in underperforming firms are prone to view dominant industry practices as the right way to follow. For instance, when most industry peers invest heavily in innovation, executives in underperforming firms are likely to view R&D investment as an effective method of achieving economic success and closing the performance gap. A strong industry norm on investing in innovation may also inspire executives in underperforming firms to form more optimistic expectations of their firms’ capability to benefit from R&D investment. Such positive expectations may further channel underperforming firms’ attention to innovation as a turnaround solution and encourage them to increase innovation expenditures in the event of performance shortfalls. This gives rise to the following hypothesis:
Hypothesis 5: Firms competing in industries with high R&D investment intensity are associated with a larger R&D investment increase in response to performance shortfalls than firms in industries with lower R&D investment intensity.
METHODS
We investigate our research questions in a longitudinal sample of high-tech firms listed in the ChiNext board of China's Shenzhen Stock Exchange. As a Chinese counterpart of Nasdaq in the US, the ChiNext board was established in 2009 to serve as a financing platform for innovative, growth-oriented firms. Firms listed on the ChiNext board are generally much smaller and younger than firms listed on the main board. Our initial sample includes all firms listed on the ChiNext board within the period of 2009–2017. We further restrict our sample to certified high-technology businesses but exclude those competing in traditional industries.[Footnote 3]
Within our sample period, there are 539 certified high-tech firms, accounting for 89.38% of firms listed on the ChiNext board by the end of 2016. Of the sample firms, 19.3% are in the software and information technology sector. Computer and electronic manufacturers account for 15.96% of the total sample, and special machinery manufacturers consist of 13.17% of our sample. Firms in other manufacturing sectors are also included, such as electrical machinery, auto and metal manufacturing, technology services, and others. We collect information on firm financials, ownership and board structures, and CEO backgrounds using the China Stock Market & Accounting Research (CSMAR) and the Wind databases. After deleting firm years with incomplete information, our final sample has 2,760 firm-year observations.
Our dependent variable is the R&D intensity ratio, which is calculated as the firm's annual R&D expenditure divided by total sales (denoted as RD/Sale). Our main independent variables are performance discrepancies relative to aspirations. Following prior research (e.g., Greve, Reference Greve2003; Kim, Finkelsteis, & Haleblian, Reference Kim, Finkelsteis and Haleblian2015; Rudy & Johnson, Reference Rudy and Johnson2016), we measure aspirations using three methods: (1) social aspirations, which refer to the performance of comparable industry peers and are calculated as the median return-on-assets ratio (ROA) of all firms listed on the ChiNext board in the focal firm's industry; (2) historical aspirations, that is, the organization's past performance calculated as the focal firm's average performance (measured by ROA) in the past two years; and (3) composite aspirations integrating both social and historical aspirations, constructed as 0.8 × social aspirations + 0.2 × historical aspirations.[Footnote 4]
As this research is concerned with firm performance below the aspiration level, performance is assessed as a spline function with separate slopes above and below zero, corresponding to firm performance equal to the respective aspiration level. We apply Under_Social, Under_Historical, and Under_Composite to measure the absolute difference in firm performance from social, historical, and composite aspirations, respectively, when firm performance is below the corresponding aspiration level. These variables are set to zero when firm performance is above the respective aspiration level. Similarly, we use Over_Social, Over_Historical, and Over_Composite to capture the absolute difference between firm performance and the respective aspiration level when firm performance is above the aspiration level and set these variables to zero otherwise.
Our model consists of five moderator variables. Tech_Function is a dummy variable, which takes the value of one when the CEO has an R&D or engineering background or has previously served as a chief technology officer (Barker & Mueller, Reference Barker and Mueller2002). Political is a dummy variable that is equal to one when the CEO used to be a government official and serves as a member of the National People's Congress or the National Committee of the Chinese People's Political Consultative Conference (Conyon et al., Reference Conyon, He and Zhou2015). Duality is a dummy variable capturing leadership structure, with one indicating that the positions of CEO and board chair are held by the same individual and zero otherwise. Industry competitiveness (denoted as Concentration) is measured using the Herfindahl–Hirschman Index of firm sales using all listed firms in the focal firm's industry. The index is calculated by squaring the market share of each firm in an industry, then summing the resulting numbers. RD_Median measures the median value of R&D intensity in the focal firm's industry. We then interact these five moderators with three underperformance measures, respectively.
We control factors at the CEO, board, firm, and industry levels that may also affect corporate innovation as identified by prior studies (e.g., Barker & Mueller, Reference Barker and Mueller2002; Belloc, Reference Belloc2012; Chen & Miller, Reference Chen and Miller2007; Miller, Reference Miller1991). The CEO level control variables include CEO_Age, CEO_Tenure, and CEO_Founder indicating the CEO's founder status, and CEO_Share capturing CEO's equity ownership. Board and ownership variables include Board_Size, Outsider_Ratio capturing the proportion of independent directors on the board, Largest_SH measuring ownership concentration using the percentage of shares held by the largest shareholder, and SOE indicating whether the firm is a state-owned-enterprise. At the firm level, we measure Slack using the current ratio computed as current assets divided by current liabilities. We measure a firm's innovation experience using the logarithmic value of accumulated R&D spending since its initial public offering year (denoted as RD_Stock). We capture Firm_Size using the logarithmic value of total assets. We use the debt-to-asset ratio (Debt_Ratio) to measure capital structure, and we also control for Firm_Age. Industry-level variables are computed using information on all listed firms, including those listed on the ChiNext board and the main board. These include industry munificence (denoted as Growth) calculated using the three-year rolling average of the industry's annual sales growth rate, industry instability (denoted as Instability) using the antilogs of the standard error of each regression slope coefficient from industry growth equations following Keats and Hitt (Reference Keats and Hitt1988), and industry profitability (denoted as Profitability) using the industry's average net profit margin. Our models also include industry dummy variables to capture other industry effects and time dummy variables to control for macro-environmental shocks. We winsorize control variables on firm characteristics by 0.025 on each end of the data distribution to reduce the influence of extreme values. Values of our variance inflation factor tests of main dependent, independent, and control variables (excluding industry and time dummy variables) range from 1.06 to 1.93, with an average of 1.45. All are below the threshold value of 2.50; thus, multicollinearity is not a major concern.
We apply a panel data fixed-effects model on firms to conduct our estimations. Differing from a pooled-ordinary least squares model that estimates between-firm variations in R&D intensity, a firm fixed-effects model captures within-firm differences, that is, changes in firms’ R&D intensity over time. This model controls for unobserved heterogeneity in firm quality to filter out time-invariant factors that may contaminate our estimates. The incorporation of firm fixed-effects also mitigates the endogeneity problem caused by unobserved firm-level confounding variables that might be simultaneously correlated with both dependent and independent variables, thus better establishing causality (Wooldridge, Reference Wooldridge2011). We also lag all independent and control variables by one period to mitigate the reverse causality problem. In addition, we account for heteroskedasticity and autocorrelation by using robust standard errors clustered at the firm level. Our model is outlined in the following equation:
where i represents the firm and t indicates the year. The dependent variable is RD/Sale. Under represents the scale of underperformance relative to aspiration levels, and Over represents the scale of overperformance. The interaction terms of Under with Tech_Function, Political, Duality, Concentration, and RD_Median measure the moderating roles of CEO functional background, political connections, leadership structure, industry competition, and industry norms on the relationship between underperformance and R&D intensity. We expect λ 1 > 0, β 1 > 0, β 2 < 0, β 3 > 0, β 4 < 0, and β 5 > 0. C represents control variables elaborated above. αi represents firm fixed effects, μt is time fixed effects, and εit is the random error term.
RESULTS
Descriptive Analysis and Univariate Analysis
Table 1 provides a descriptive analysis of our key variables. The average R&D intensity in our sample period is 0.071. The average scale of social underperformance is 0.018, historical underperformance is 0.026, and composite underperformance is 0.017, which indicates that underperforming firms’ ROA is 0.018, 0.026, and 0.017 lower than their respective aspiration levels. In contrast, the average scale of social overperformance is 0.019, historical overperformance is 0.014, and composite overperformance is 0.017. Table 1 also reveals that 48.2% of CEOs have a technical functional background, and 22.1% of CEOs have political connections. In addition, 44.1% of firms have a combined leadership position. The average industry concentration ratio is 0.075, and the industry median R&D intensity is 0.058. Moreover, an average CEO in our sample is about 48 years old with 5 years of experience in the position, and 50.9% of CEOs are founders of their firms. This pattern is consistent with the goal of the ChiNext board to attract young entrepreneurial firms. An average board in our sample consists of eight members, and 38% of these members are independent directors. Only 2.2% of firms are state-owned enterprises, which is in sharp contrast with firms listed on the main board, where SOEs account for nearly half of listed firms. The largest shareholder on average owns 32.24% of outstanding shares. Furthermore, an average firm in our sample has a current ratio (i.e., slack) of 5.358, an average R&D stock of 4.066, a debt-to-asset ratio of 0.267, an average size of 6.155 measured by the logarithm of total assets, and is 15 years old. Finally, the average industry growth rate is 0.23, the average industry instability measure is 0.959, and the average industry profit margin is 0.064.
Table 1. Descriptive analysis of key variables
Table 2 presents the correlation matrix of key dependent, independent, and control variables. We find that all three underperformance measures are positively associated with R&D intensity, which supports our baseline hypothesis. We also notice that CEOs with R&D or engineering backgrounds are associated with higher R&D intensity, and politically connected CEOs are associated with lower R&D intensity. In addition, both leadership duality and industry median R&D intensity are positively correlated with R&D intensity, whereas higher industry concentration is associated with lower R&D investment intensity. We also identify a positive relationship between CEO founder status and R&D investment intensity. We find that firms with younger CEOs and more independent boards are associated with higher R&D investment intensity, while board size and ownership concentration have a negative relationship with R&D investment intensity. In addition, organizational slack, R&D investment stock, and the industry growth rate are all positively correlated with R&D investment intensity, and firm size and liquidity ratio are negatively related to R&D investment intensity. Finally, we notice that overperformance is associated with lower R&D investment intensity. Overall, these results are consistent with empirical findings of previous innovation literature.
Table 2. Correlation matrix of key variables
Note: *p < 0.05.
Multivariable Analysis
We report empirical results obtained from fixed-effects models in Tables 3–5. Table 3 captures underperformance relative to social aspirations, Table 4 reports results related to historical aspirations, and Table 5 demonstrates results on blended aspirations compiled using both social and historical benchmarks. In all three tables, Column 1 includes only independent and control variables without interaction terms to serve as the baseline. Column 2 adds the interaction of underperformance with Tech_Function to test Hypothesis 1 (H1), and Column 3 adds the interaction term with Political to test Hypothesis 2 (H2). In Column 4, the interaction with Duality is added to test Hypothesis 3 (H3), and Column 5 adds the interaction with Concentration to examine Hypothesis 4 (H4). Column 6 adds the interaction with RD_Median to examine Hypothesis 5 (H5), and finally, all five interaction variables are included together in Column 7.
Table 3. Effect of social underperformance on R&D investment
Notes: Underperformance is measured relative to social aspirations. Fixed-effects models on firms are applied. All independent and control variables are lagged for one period. Robust standard errors are reported in parenthesis. *p < 0.10; **p < 0.05; ***p < 0.01.
Table 4. Effect of historical underperformance on R&D investment
Notes: Underperformance is measured relative to historical aspirations. Fixed-effects models on firms are applied. All independent and control variables are lagged for one period. Robust standard errors are reported in parenthesis. *p < 0.10; **p < 0.05; ***p < 0.01.
Table 5. Effect of composite underperformance on R&D investment
Notes: Underperformance is measured relative to composite aspirations including both social and historical benchmarks. Fixed-effects models on firms are applied. All independent and control variables are lagged for one period. Robust standard errors are reported in parenthesis.*p < 0.10; **p < 0.05; ***p < 0.01.
First, we observe a consistently statistically significant relationship between underperformance magnitude and R&D investment intensity in the baseline model, no matter whether underperformance is measured relative to social, historical, or composite aspirations. This result indicates that underperforming firms increase innovation investment in response to performance shortfalls, which supports our baseline hypothesis. In addition, we find that underperforming firms whose CEOs have a technical background are associated with an even larger increase in R&D investment intensity than those firms with CEOs of other backgrounds in Tables 3–5. Thus, H1 is consistently supported across models and measures.
We also find that underperforming firms whose CEOs have political ties are associated with a decrease in R&D investment intensity when firm performance is below the aspiration level. These results hold when underperformance is measured relative to social, historical, and composite aspirations. Taking social underperformance as an example, the coefficient of Under_Social is 0.058, and the coefficient of the interaction term Under*Political is −0.166. This result indicates that one unit increase in the magnitude of underperformance relative to the social benchmark results in a 0.108 (0.058–0.166) unit decline in the R&D intensity of firms with politically connected CEOs. We thus find consistent support for H2 on the negative moderating role of CEO political connections. In addition, we document that underperforming firms with a combined leadership structure are associated with an even larger increase in R&D intensity than firms with separate CEO and chairperson positions across all models and measures, which confirms H3 on the moderating role of leadership structure.
Moreover, we find that social underperforming firms in competitive industries increase their R&D investment to a larger degree than those in more concentrated industries as predicted by H4. However, historical underperforming firms are associated with a larger increase in R&D investment when they compete in concentrated industries than in more competitive industries, while no significant relationship is identified when underperformance is measured using the blended measure. Thus, H4 is not consistently supported. As Kim et al. (Reference Kim, Finkelsteis and Haleblian2015) suggest, historical and social aspirations may lead to distinct managerial behavior because their underlying mechanisms and benchmarking processes vary with each other. In this case, industry competition prompts social underperforming firms to increase their R&D investment to a larger degree, while deterring historical underperforming firms from doing so.
Finally, we find that underperforming firms increase their R&D investment even more when the industry's median R&D investment intensity is higher if underperformance is measured against social or composite aspirations. We do not identify this effect in the event of historical underperformance. These results indicate that firms imitate their industry peers’ innovation investment strategies when their performance falls below the level of peers’. However, a strong industry norm to invest in innovation has scant influence on firm behavior when firm performance falls below historical aspirations. Therefore, H5 is only partially supported.
Sensitivity Analyses
We utilize a panel data fixed-effects model for our main analyses. In that some sample firms may have minimal R&D investment in given years, our dependent variable, R&D intensity, is affected by zeros on the left side of sample distributions. We thus implement a random-effects Tobit model, also known as a censored regression model, to perform sensitivity analyses.[Footnote 5] We report results of this alternative model in Columns 1–3 of Table 6, with Column 1 on social underperformance, Column 2 on historical underperformance, and Column 3 showing composite underperformance. Our dependent variable, RD/Sale, is a proportion variable between 0 and 1, with the minimum value being 0 and the maximum value being 0.92. We next apply a fractional response model, more specifically, the fractional logit model recommended by Wooldridge (Reference Wooldridge2011) to conduct sensitivity analysis.[Footnote 6] Results of the fractional logit model are reported in Columns 4, 5, and 6 of Table 6, representing social, historical, and composite underperformance, respectively.
Table 6. Effect of underperformance on R&D investment sensitivity analyses
Notes: Random-effects Tobit models are applied in Columns 1–3. Fractional logit models are applied in Columns 4–6. All independent and control variables are lagged for one year. Robust standard errors are reported in parenthesis. *p < 0.10; **p < 0.05; ***p < 0.01.
We find consistent support that underperforming firms are associated with an increase in R&D expenditure, which confirms our baseline hypothesis. In addition, we document that the CEO's functional background positively moderates the relationship between underperformance and R&D investment change in both the Tobit model and the fractional logit model, which confirms H1. We also find additional support for H2 that underperforming firms with politically connected CEOs are associated with a decrease in R&D investment when the Tobit model is applied, but not when the fractional logic model is utilized. Moreover, Table 6 verifies H3 that underperforming firms with a combined leadership structure are associated with a larger increase in R&D intensity than firms with a separate leadership structure. We find inconsistent support across models for the moderating effect of industry competition. Finally, results of the fractional logit models provide additional support for H5 that firms competing in industries with stronger norms to invest in R&D are associated with even larger increases in R&D investment when performance falls below aspirations.
In our main analyses, we construct social comparison groups using all other firms listed on the ChiNext board and competing in the same industry. Firms listed on the ChiNext board are close competitors owing to their similar size, history, and listed stock exchanges. However, focal firms may benchmark against all listed firms in the same industry, including larger and older firms listed on the main board. We also reconstruct our social underperformance and overperformance measures using this broader social comparison group and reperform all analyses. Our unreported results reveal a statistically significant positive relationship between this reconstructed social underperformance measure and changes in R&D investment intensity and confirm moderating effects as predicted by H1, H2, H3, and H5 as well.
DISCUSSION
This study examines how performance shortfalls may prompt underperforming firms to adjust R&D investment intensity. We explore how managerial cognition, as shaped by CEOs’ experiences, connections, positions, and industry contexts, affects executives’ perception of innovation as a valid solution to address underperformance problems and consequently moderates the relationship between negative performance feedback and innovation investment decisions. We conduct our analysis using a longitudinal sample of listed Chinese high-tech entrepreneurial firms. Our results indicate that when firm performance falls below the aspiration level, entrepreneurial firms will increase R&D investment intensity. Performance shortfalls are associated with an even larger increase in R&D investment when CEOs of underperforming firms have an R&D or engineering background, serve simultaneously as the board chair, or are not politically connected. We also find that industry environment matters, although the effect varies with aspiration types. Social underperformance is associated with a larger increase in R&D investment if the focal firm is competing in a more competitive industry or the industry's average R&D investment level is higher. We conclude that dispositional and situational factors shape managerial cognition, influencing executives’ allocation of attention to innovation as a feasible solution to close performance gaps, and subsequently affecting firms’ responses to performance shortfalls.
Managerial scholars have continuously called for a better understanding of how organizations learn from other organizations and from their own experiences, and how performance feedback drives problemistic search and organizational change (Gavetti et al., Reference Gavetti, Greve, Levinthal and Ocasio2012). Firms’ responses to performance feedback are heavily influenced by the way in which their key decision-makers interpret this information and act upon it. This study highlights managerial cognition and attention allocation as important mechanisms shaping the direction and magnitude of organizational changes. Following insights of the attention-based view on the focus of attention, structural distribution of attention, and situated attention, we predict and confirm that a manager's allocation of attention to innovation is influenced by a wide array of dispositional and situational factors. Heterogeneity in executives’ functional backgrounds and social connections, diversity in firm leadership structure, and variations in industry environments all influence decision-makers’ attention allocation patterns and their evaluation of corporate innovation as a solution to close performance gaps and consequently result in distinct firm responses to performance feedback. As such, our article extends the managerial cognition and attention-based view literature by explicitly modeling the moderating role of managerial cognition on the relationship between performance aspirations and firm actions. We argue that a firm's actions are guided not only by its aspirations and performance feedback, which affects executives’ attention to specific issues, but also by managers’ allocation of attention to certain solutions to address performance problems. This cognitive angle therefore enables us to enrich the performance feedback theory and the organizational learning literature by examining how the cognition of corporate decision-makers influences their capabilities to learn from past failures and guides their future innovation decisions.
Our results also demonstrate that industry environment has different effects on firm responses to performance shortfalls dependent on whether aspiration levels are constructed based on social peers or historical performance. These two types of aspirations are interpreted differently by managers and are filtered through dissimilar cognitive and organizational processes because of their distinct underlying benchmarks (Kim et al., Reference Kim, Finkelsteis and Haleblian2015). Social aspirations reveal how firms perform in the eyes of the market and stakeholders. Social underperformance often signals organizational problems to internal and external entities. When competing in innovation is the dominant industry strategy, imitating this prevailing strategy by increasing investment in innovation helps firms falling behind social aspirations quickly gain legitimacy. In contrast, historical aspirations closely reflect managerial capabilities and firm resources. Historical underperformance results in a downward adjustment in the assessment of firm capability and prompts managers to update their assumptions on the effectiveness of current strategies and practices. In this case, even if innovation is the dominant industry strategy, firms questioning their capabilities may not necessarily view these resource-consuming strategies as a valid solution to close performance gaps. Our study thus indicates that the allocation of managerial attention not only varies by dispositional and situational factors but also depends on the construction of reference points.
Our study also adds to the growing literature on corporate innovation by providing a systematic framework explaining how dispositional and situational factors at the executive, firm, and industry levels interact to shape managerial cognition, direct firm attention, and ultimately affect firm decisions to invest in innovation. This cross-level framework, integrated under the umbrella of the attention-based view, augments prior innovation research that has examined influences of CEO and executive characteristics, firm endowments, task environment, and institutional environment in isolation to strengthen the understanding of the interplay of micro-, meso-, and macro-level factors shaping corporate innovation decisions.
Our investigation also possesses crucial practical implications. Feedback in the form of relative financial performance is one of the most important sources of information for executives to evaluate the efficacy of current firm strategy and to decide the need for future strategic change (Schumacher, Keck, & Tang, Reference Schumacher, Keck and Tang2020). It is thus important for executives to clearly discern how their interpretation of performance feedback and subsequent strategic choices are affected by their functional background, social connections, positions in firms, and industry environment. Realizing the influence of these dispositional and situational factors on decision-making will enable executives to better overcome their cognitive constraints and to make sound strategic decisions for their firms. In addition, our findings suggest that politically connected CEOs tend to respond to underperformance by decreasing instead of increasing innovation investment. Thus, overcoming the tendency to excessively rely on political patronages for assistance will help firms reconsider innovation as a long-term strategy for performance improvement and competitive advantage.
Limitations and Future Research Directions
Our work is also subject to several limitations that warrant future research. First, we restrict our sample to high-tech entrepreneurial firms. These firms usually adopt a differentiation strategy and gain competitive advantage through differentiated products or services. Thus, investing in innovation is a salient strategic choice that attracts the attention of corporate executives. However, results of our article may not hold for firms competing in traditional industries and those deploying a cost leadership strategy. Because R&D is an expenditure item that directly increases operation costs in the short run and is risky with uncertain long-term payoffs, underperforming firms facing financial distress may utilize cost-cutting and resource conservation instead of boosting R&D investment to improve performance (Bromiley & Washburn, Reference Bromiley and Washburn2011; Greve, Reference Greve2010). Future studies may adopt a broader sample to examine how business strategy and industry environment affect changes in firms’ innovation strategy in case of performance shortfalls. Second, this study constructs aspiration levels using the additive model that assumes decision-makers consider both self-referents and social referents separately and in tandem when assessing performance (Greve & Gaba, Reference Greve, Gaba, Argote and Levine2017). We do not investigate how inconsistent negative performance feedback between social and historical aspirations affects firms’ R&D investment decisions. Neither do we explore how managers may put different weights on these aspirations and switch their attention from one reference to the other over time. Future research can further explore how firms set, select, and weight different aspirations levels, and how firms’ allocation of attention to different reference points may impact organizational search and changes. Third, our article measures R&D investment in aggregate but does not differentiate between explorative and exploitative innovation projects. Tushman and O'Reilly (Reference Tushman and O'Reilly1996) suggest that organizational ambidexterity – the ability to simultaneously pursue both incremental and revolutionary innovation – is essential for firms’ competitive advantage. Investigating how managerial cognition affects firms’ decisions to invest in exploitative or explorative projects will also be a fruitful future research avenue. Finally, our study examines how leadership structure affects decision-makers’ attention allocation and their corresponding responses to performance feedback. Future studies can explore the influence of other structural factors on decision-makers’ setting of aspirations, interpretation of performance feedback, allocation of attention, and subsequent action choices.
CONCLUSION
In summary, our article highlights the essential role of managerial cognition and attention allocation in the linkage between organizational aspirations and actions. We postulate and find that firms respond to performance shortfalls by increasing innovation investment when executive attention is allocated to innovation as a valid performance-enhancing solution. Decision-makers’ allocation of attention is influenced by their prior experience, firm leadership structure, and industry context. Our study enriches the performance feedback theory by explaining and investigating the influence of managerial cognition on underperforming firms’ search trajectory and action choices. We trust that this work provides vital practical implications for how performance feedback-based learning can help organizations gain and sustain competitive advantage through innovation.
