Affective states and a firm’s performance: the mediating role of dynamic managerial capabilities

Abstract

Purpose

Extant research supports the importance of dynamic managerial capabilities in capturing managers’ individual roles in organisations’ adjustments to change. This paper develops a multidimensional scale for measuring dynamic managerial capabilities consisting of sensing, seizing and reconfiguration capacities that mediate between managers’ affective states and their firms’ performance.

Design/methodology/approach

The scale is validated in a survey-based study among 204 managers in companies in the United States of America (USA). We applied a multiple regression model (a triple mediation) using each of DMCs’ three dimensions to test the effects of managers’ affective states on their firms’ performance.

Findings

The multidimensional construct of DMCs adds about 15 % of variance explained to a firm’s performance, as perceived by its managers. So managers’ affective states do have an impact on DMCs and, later, on their firms’ performance.

Research limitations/implications

We show the impact of negative and positive affect on DMCs. We also show that DMCs’ three dimensions should be treated in a formative manner that advances discussion on DMCs and their role in a firm’s performance.

Practical implications

Understanding managers’ affective states helps incorporate “hot cognition” into firms’ strategising processes. Although both positive and negative emotions can be helpful, depending on the situation, positive affect is generally more valuable than negative affect as it relates to a firm’s performance.

Originality/value

Our work proposes measuring DMCs based on Teece’s (2007) disaggregation of DMCs into sensing, seizing and reconfiguration capacities. We approach each of these dimensions separately and show that managers’ affective states influence each dimension differently.

Citation

Bogodistov, Y. and Schmidt, S. (2024), "Affective states and a firm’s performance: the mediating role of dynamic managerial capabilities", Baltic Journal of Management, Vol. 19 No. 6, pp. 111-132. https://doi.org/10.1108/BJM-09-2023-0352

1. Introduction

Firms that operate in highly dynamic environments require dynamic capability (DC), that is, the ability to sense threats and opportunities (sensing), seize an opportunity (seizing), and reconfigure their resource bases (reconfiguration) accordingly (Barreto, 2010; Teece, 2007). However, not only organisational capabilities but also managers’ individual capabilities–so-called dynamic managerial capabilities (DMCs)–explain variations in corporate strategies in turbulent times (Adner and Helfat, 2003). According to Ambrosini and Altintas (2019, p. 1), DMCs “are concerned with the role of managers in refreshing and transforming the resource base of the firm so that it maintains and develops its competitive advantage and performance.”

However, environmental turbulence might be an issue for individuals who are prone to be influenced by their affective states when they make decisions (Slovic et al., 2005). When facing novel challenges, managers may be afraid because of threats to their business, angry about encountering obstacles in their way, irritated by noise in the information flow, or distressed by their employees’ loss of faith in their companies. Both positive and negative affective states play a role in managerial decision-making (Hodgkinson and Healey, 2011; Huy and Zott, 2019), but we lack clarity regarding how this influence affects firms’ performance, that is, why and how individual-level affective states lead to a drop or increase in a firm’s performance.

As DMCs are part of the broader DC construct, DMCs should inherit the characteristics of the superordinate DC categories of sensing, seizing, and reconfiguration. These characteristics refer to managers’ individual capabilities to sense opportunities and threats, seize and elaborate on them, and manage resource reconfigurations in response, thus shaping their firms’ performance. As Teece (2007, p. 1320) observes, “The ambition of the dynamic capabilities framework is nothing less than to explain the sources of enterprise-level competitive advantage over time and provide guidance to managers for avoiding the zero profit condition that results when homogeneous firms compete in perfectly competitive markets” [italics added]. Firm-level DCs offer guidance, but it is individual managers who pay attention to or ignore threats, pursue opportunities, and manage the incorporation of opportunities into their firm’s strategy. However, these managers are prone to emotional biases that might affect their capacity to sense opportunities and threats, to seize them, and to reconfigure their firms’ resource bases in response. Are DMCs the mechanism that explains the impact of individual-level positive or negative affective states on firm performance?

While we know that DCs are dynamic in their nature (Teece, 2007) and are probabilistically (positively) related to a firm’s performance (Teece, 2018), and while we know the microfoundations (i.e. the mechanisms of functioning) of organisational capabilities like routines and heuristics (Bingham and Eisenhardt, 2011; Bingham and Haleblian, 2012; Eisenhardt and Martin, 2000; Felin et al., 2007, 2012; Winter, 2013), we lack knowledge about the microfoundations of individual-level DMCs. Helfat and Peteraf (2015) offer managerial cognitive capabilities as a microfoundation of DMCs, emphasising the roles of perception and attention in managerial sensing, problem-solving and reasoning in managerial seizing, and language and communication in managers’ reconfiguration capacity. Levine et al. (2017) support the same rational cognitive approach and stress the importance of analytical skills and strategic intelligence for performance. However, both approaches assume rationality and cold cognition, ignoring affective states that must have an impact on managerial decision-making (Ashton-James and Ashkanasy, 2008; Healey and Hodgkinson, 2017; Huy and Zott, 2019).

Research on D(M)Cs focuses on the role of rationality and “cold cognition” over that of emotions and other affective states (Healey and Hodgkinson, 2017; Hodgkinson and Healey, 2011). Although Helfat and Peteraf (2015) argue that affective states play an essential role in managers’ social cognition and language and communication capabilities, other scholars see affect as a moderator of rational decision-making, rather than an independent antecedent. Hodgkinson and Healey (2011) propose the idea of “hot cognition”, a mode of information processing when a manager’s affective state may substitute for or shift logical arguments and become the main predictor of a decision.

In this work, we argue that both positive and negative affective states influence DMCs, leading to increases or decreases in firm performance. We address the possibility that DMCs are a mediator that explains why positive or negative affect, an individual-level concept, leads to an increase or decline in firm performance. Using a scale for DMCs that we developed, we investigate how DMCs are restricted or broadened by emotions that managers experience.

We use an online survey of managers from 204 firms that are based in the US and that operate in various industries to find the answer to our research question: Is affective states’ influence on firm performance mediated by managers’ sensing, seizing, and reconfiguration capabilities, that is, their DMCs? Our study shows that positive affect has a generally positive effect on DMCs and, ultimately, on performance. However, the influence of affective states is not straightforward, as negative affect is shown to be bad for a firm’s performance but good for the reconfiguration dimension of DMC. We address this finding in our Discussion and Managerial Implications sections.

2. Theoretical background

While the DC concept can be used to explain firm success in dynamic environments (Teece et al., 1997), the DMC concept focuses on the individual manager in a firm and his or her role in the firm’s success (Sirmon and Hitt, 2009). Adner and Helfat (2003, p. 1012) define DMCs as “capabilities with which managers build, integrate, and reconfigure organisational resources and competencies” and approach them through three pillars: managerial human capital, managerial social capital, and managerial cognition. However, their approach finds less support from the academic community than the sequential (also chain or procedural) logic from Teece (2007) (Barreto, 2010; Bogodistov and Moormann, 2024; Helfat and Peteraf, 2009; Schilke et al., 2017), which allows managerial actions, not their potential (e.g. the number of social connections–social capital–or type of education–human capital), to be in focus. In the end, capabilities are “complex, team-based productive activities” (Grant, 1996, p. 116 [italics added]), while a potential may not be activated (Barreto, 2010). This approach also distinguishes DMCs from luck and non-repetitive ad-hoc problem-solving (Winter, 2003).

In their search for clarity regarding the nature of DC and for insights on why DCs influence a firm’s outcomes (i.e. microfoundations), scholars focus either on organisation-level mechanisms like routines (Felin et al., 2012; Salvato and Rerup, 2018) or organisation-wide heuristics (Bingham and Haleblian, 2012; Eisenhardt and Martin, 2000), or on individual-level predictors like individual values, mindsets, or emotions (Scheuer and Thaler, 2023). These individual-level factors, like affective states, can alter managerial decisions, mobilise teams (Huy and Zott, 2019), or even shape organisation members’ behaviour in their firms (Schilke et al., 2017). The impact of affective states in strategy and strategizing is often ignored or stigmatised (Hodgkinson and Healey, 2011), but “eradicating emotional influences from the strategy process is not only infeasible, it is also undesirable” (Healey and Hodgkinson, 2017, p. 109).

Still, the underlying mechanism in how affect influences DMC is not well understood. DMC is a complex construct that incorporates managers’ human and social capital and cognitive capabilities like mental models, perception, attention, reasoning, and problem-solving (Adner and Helfat, 2003; Helfat and Peteraf, 2015). Emotions are a complex evolutionary phenomenon represented by primary emotions and secondary emotions, where the latter incorporate cognitive processes (Damasio, 2001) into the former. Therefore, the same situation may evoke different affective responses, as a manager may interpret new technology as a threat (resulting in fear), as an obstacle (resulting in anger), or as an opportunity (resulting in happiness). Although we know how emotions appear and which mechanisms underlie each one (Damasio, 2008; Ekman, 2007; Frijda, 1993; Frijda and Parrott, 2011), research on emotions in the organisational context is lacking. Research on affect in organisational settings is one of the research avenues scholars of DCs most frequently demand (Scheuer and Thaler, 2023).

2.1 Dynamic managerial capabilities

The idea of capturing the essence of DMCs is not new. While several researchers approach DMC through the prisms of managerial human and social capital and managerial cognition (Adner and Helfat, 2003; Corrêa et al., 2019), we apply Teece’s (2007) more popular framework, which looks at DMC through the prism of managerial sensing, seizing, and reconfiguration capacities [1]. Reviewing these dimensions may suggest that the dimensions appear sequentially, which speaks to the formative logic of the DMC construct and a mediation relationship within the construct itself that may explain the drop (or growth) in a firm’s performance as a chain reaction triggered by managerial sensing and leading to shifts in reconfiguration via managerial seizing.

The logic of a DC’s sequential appearance is largely supported on a conceptual level (Barreto, 2010; Wilden et al., 2018) and by empirical investigations (Li and Liu, 2014; Wilden et al., 2013). Moreover, this logic has an implication for empirical research: A manager needs to sense an opportunity or threat before it can be incorporated into decision-making. Similarly, the manager may not know how to reconfigure the firm’s resource base before having made the decision. Consequently, the outputs of sensing become inputs for seizing, and the outputs of seizing become inputs for reconfiguration. For instance, if the input for a sensing capability is the (limitless) opportunities and threats from the market, its output is a set of opportunities that the manager considers to be important for the firm. Sensing capacity processes the information, via the mechanisms of attention and perception, into a set of decision options (Helfat and Peteraf, 2015) that are processed using a seizing capacity, and the output–the most promising opportunity–appears. This opportunity is then addressed by means of reconfiguration, whereby a manager applies social and communication skills to reconfigure the firm’s resource base (Helfat and Peteraf, 2015).

Our approach to sensing, seizing, and reconfiguration has two key implications. First, we see the dimensions that underlie a DMC as formative, rather than reflective. In a reflective measurement, the causality flows from the latent construct (e.g. DMC) to its indicators (e.g. sensing, seizing, and reconfiguration), while in a formative logic, the causality is reversed (Coltman et al., 2008). For DMC, “dropping an indicator would be similar to dropping a part of the construct” (Freeze and Raschke, 2007, p. 1483), supporting the formative approach. The second key implication of our approach is that the overall output of a dynamic managerial capability may be the result of a complex mediation process, whereby the dimensions of sensing, seizing, and reconfiguration are not even correlated, as one can imagine would be the case for a firm that has significant capabilities to notice opportunities and threats while also having problems making decisions. We depict this structure in our research model (Figure 1).

Following this model, affective states are particularly important at the beginning of the process, as if affective states can weaken or strengthen the sensing capacity, all of the following capacities (and the whole DMC) are influenced. Hodgkinson and Healey (2011) stress that “hot cognition” in sensing might help to avoid cognitive blind spots and strategic inertia, but the other managerial capacities may be influenced by affective states as well.

2.2 Affect and a manager’s ability to handle change

An “affective state” is a broad term that incorporates concepts related to moods, feelings, and emotions. These concepts can be categorised based on intensity, as a mood is less intense than an emotion; duration, as an emotion is experienced for a shorter time than a mood, and a feeling lasts somewhere in the middle; and diffuseness, as an emotion is related to a certain stimulus such that, after the stimulus is forgotten, it becomes a feeling and may later may turn into a mood (Frijda, 1993). In this paper, we use the terms “affective states” and “emotions” interchangeably, whereby our focus is closer to long-term moods than short-lived emotions.

Affective states are an evolutionary mechanism that humans and animals developed to cope with the complexity of the environment (Barrett, 2016; Ekman, 2009). Affect refers to uncontrolled (primary emotions) or controlled (secondary emotions, involving cognitive functions) ways to process information from the environment and prepare the body for a response (Damasio, 2001, 2008). Each discrete emotion has its underlying theme (e.g. threat or danger for fear, loss for sadness, or self-blame for guilt) (Smith and Lazarus, 1993) and its function (e.g. fear causes behavioural inhibition, readiness to flee, or necessity to freeze, whereas anger causes behavioural activation in the form of readiness to engage with an obstacle or fight an enemy (Carver, 2006; Carver and White, 1994). The uncontrolled response is not always linear for evolutionary reasons: Fear causes some people to freeze and others to flee and all to fight when freezing and flight are not options. Indeed, animal observations show that “[i]f escape was possible, the animal tried to flee; if not, it tried to fight” (Misslin, 2003, p. 58).

Understanding affective states is key in strategic management, as individual-level affective states can lead to cognitive blindspots that restrict managerial sensing capacity (Hodgkinson and Healey, 2011), positively or negatively influence decision-making (and, thus, seizing capacity) (Ashton-James and Ashkanasy, 2008), or spark managers’ to engage in proactive behaviour (Lebel, 2017). Hodgkinson and Healey (2017, p. 129) address emotions in their notion of “hot cognition” as an “approach to strategic adaptation, highlighting the significance of emotion and related affective processes.” They suggest that “the fundamental capabilities of sensing, seizing, and transforming each require firms to harness the cognitive and emotional capacities” (Hodgkinson and Healey, 2011, p. 1500). With their approach, the authors emphasise that affect should be introduced into a firm’s strategy via tools and processes that aim to change a manager’s mental model. A strategy that is in congruence with emotional mechanisms can enhance a firm’s DMCs. Thus, we may link the firm-level strategising processes with the individual-level affect that is inseparable from an individual, as an individual’s actions are predisposed by affect, cognition, and habit (Dewey, 1922). These action-related predispositions play a key role in organisational heuristics, routines, and capabilities (Bogodistov and Moormann, 2019; Cohen, 2007; Winter, 2013).

Organisational change is accompanied by a range of emotions from anger to happiness (Helpap and Bekmeier-Feuerhahn, 2016), depending on the individual’s interpretation of the event. As organisational change is multifaceted, different aspects of it can elicit different emotions.

2.3 Research model

These theoretical arguments support the development of a research model. The influences of positive and negative affect on managerial sensing might differ. Immediate emotions transform into a positive or a negative mood, which shifts a manager’s perceptions and influences his or her sensing capacity (Helfat and Peteraf, 2015). For instance, positive moods are associated with creativity and cognitive flexibility (Estrada et al., 1994; Isen et al., 1987), whereas negative moods may result in avoidance behaviour (Frijda et al., 1989), lowering the manager’s willingness to search for new opportunities (or threats). This shift could reduce the scope of the manager’s search, change his or her perceptual schema, or decrease the speed of learning (Zahra and George, 2002).

As for managerial seizing, negative affect may restrict managers’ ability to assimilate and apply new knowledge (Alexiou et al., 2019) or may lead to over-optimistic risk assessments, risk-seeking behaviour in the event of anger, decreased selling and increased buying prices in the event of sadness, or pessimistic situation appraisals in the event of fear. However, positive affect may lead to motivated actions and promote helpful behaviour (Ashton-James and Ashkanasy, 2008). Hodgkinson and Healey (2011, p. 1506) argue that, to be able to seize sensed opportunities, firms must be able to evaluate them in a progressive and forward-looking manner and to “unlock dysfunctional fixations with existing strategies.” Positive emotions may help overcome such fixations and encourage the cognitive flexibility (Johnson et al., 2010) that can be advantageous for problem-solving, one of the managerial cognitive capabilities that underlie seizing (Helfat and Peteraf, 2015). In short, negative emotions are likely to induce a more rigorous information search, whereas positive emotions are likely to induce quick and flexible decision-making (Spering et al., 2005).

As for managerial reconfiguration, Alexiou et al. (2019, p. 165) show that “positive employee states are not only important in their own right but are also tied to desirable organisational outcomes,” and Lebel (2017) shows that both fear and anger drive proactive behaviour and may help firms improve their functional outcomes. Similarly, individuals in a negative affective state tend to speak proactively (Lebel, 2016), which is of significant value during an implementation phase like reconfiguration, when effective communication, loyalty-building, and commitment are required (Schweiger and DeNisi, 1991; Teece, 2007). Moreover, because “the boundaries of the enterprise need to be artfully contoured” (Teece, 2007, p. 1332) during the reconfiguration of a firm’s assets, negative emotions like anger may help to overcome organisational obstacles (Ashton-James and Ashkanasy, 2008; Bogodistov, 2015). At the same time, positive affect (if not overly intense) may reduce interpersonal conflict and increase the tendency to cooperate (Ashton-James and Ashkanasy, 2008; Isen and Baron, 1991). As communication and social cognition are related to reconfiguration (Helfat and Peteraf, 2015), we expect positive affect to be helpful in managers’ endeavours to change their firms’ resource bases.

Overall, then, affective states may shift the “ability to recognize the value of new information, assimilate it, and apply it to commercial ends” (Cohen and Levinthal, 1990, p. 128), that is, the ability to influence a firm’s performance indirectly via managerial sensing, seizing, and reconfiguration (i.e. DMC). Therefore, we hypothesise:

H1.

High levels of positive affect experienced by managers have a positive influence on their (a) sensing, (b) seizing, and (c) reconfiguration capacities and, (d) via DMC, on their firms’ performance.

H2.

High levels of negative affect experienced by managers have a negative influence on their (a) sensing and (b) seizing capacities but (c) a positive influence on their reconfiguration capacity and, (d) via DMC, on their firms’ performance.

Figure 1 depicts our research model as a chain reaction, that is, as an influence of affect on sensing capacity that echoes in the firm’s performance, as sensing affects managers’ seizing and reconfiguration capacities.

3. Methodology

3.1 Data collection and sample

We tested our research model empirically by conducting an online survey among managers of 204 firms in various industries that were based in the US. The data were collected in 2018 in three waves (May, June, and July). We found no significant differences in reported DMC between data collection periods (ANOVA: p > 0.05). Over a period of three weeks, a survey agency contacted firms via phone before sending out email invitations to participate. The invitations contained a link to the online survey. The final sample consists of responses from 204 participants who held managerial positions. The sample includes a diverse set of managers with regard to individual characteristics. For instance, the mean age is 31.39 years (SD = 8.14), 48% (n = 98) of the sample are female, and the mean length of current experience with the current firm is 12.74 years (SD = 10.46). The participants worked in managerial positions in a variety of functions. For example, 76 participants held positions on their firm’s management board, 46 held IT positions, 16 held positions in production, 13 were in marketing and sales, and 13 were in finance and controlling. Ninety-nine participants worked in firms with ≥250 employees and 141 worked in SMEs with <250 employees. Thirty-four respondents came from the IT sphere, 23 from professional services (consulting, auditing, etc.), 19 from construction, 18 from financial and insurance services, 17 from the production industry, and others from other industries. Table S1 in the Supplementary Material provides an overview of the sample.

3.2 Measures

3.2.1 Independent variables

Since we examine affective states in our research model, we used the Positive and Negative Affect Scale (PANAS) (Watson et al., 1988), a reliable, established measure for positive and negative affect that has been used in the context of firms’ innovative capabilities (Mitchell et al., 2021), entrepreneurship (Baron and Tang, 2011), and individuals’ cognitive capabilities (Barber et al., 2020). The 20 items cover two second-order constructs: positive affect (interested, excited, strong, enthusiastic, proud, alert, inspired, determined, attentive, active) and negative affect (distressed, upset, guilty, scared, hostile, irritable, ashamed, nervous, jittery, afraid) and are measured using a seven-point Likert scale ranging from “not at all” to “extremely”. The PANAS scale does not focus on an immediate emotion but on feelings that appeared during the previous year, an approach that is based on the Affective Events theory’s view that, in general, feelings accumulate and shape an individual’s behaviour (Weiss and Cropanzano, 1996).

3.2.2 Dependent variables

We used conceptual considerations of DMCs and the scale development method to adapt and extend existing scales from the literature on DCs. As a basis, we used the established scale from Wilden et al. (2013) and Li and Liu (2014), who measure DCs using surveys. We rephrased items to fit the concept of DMCs (e.g. changing “We invest in finding solutions for our customers” from (Wilden et al., 2013, p. 83) to “I invest time and effort in finding solutions for our customers”). We also added items that fit our research purpose and validated them by consulting with several scholars who work with the DC concept. Their feedback led to the final version of the 24-item scale. Respondents assessed all items using a seven-point Likert scale that ranged from “strongly disagree” to “strongly agree” for sensing and from “never” to “very often” for seizing and reconfiguration.

After we collected the data, we assessed factor loadings and cross-loadings. We included 16 items that had high factor loadings (>0.750) and low cross-loadings on seminal sub-constructs (<0.350) to ensure that the final scale consisted of items with high reliability (Cronbach’s alpha >0.880; composite reliability >0.880). Table 4 presents the final list of items, and Table 1, Table 2, and Table 3 present the reliability and validity measures. Table 4 shows the results from the maximum likelihood test, as AMOS uses this procedure to calculate model fit.

We used Vorhies and Morgan’s (2005) scale to measure firm performance. The scale measures the performance construct via items related to customer satisfaction, market effectiveness, and “current (anticipated) profitability using seven-point Likert scales that range from “much worse than competitors” to “much better than competitors”. The three reflective dimensions were aggregated using a maximum-likelihood weighting procedure.

3.2.3 Control variables

We controlled for firm size since it may affect the formation of a DC (Bogodistov and Wohlgemuth, 2017). Since DMCs are individual-level constructs, we also included controls for gender and age. As Adner and Helfat (2003) contend that DMCs are formed in part by managers’ human and social capital, we controlled for education as a proxy for human capital and for work experience in the firm as a proxy for social capital. We coded the respondents’ education level ordinally as 1 for BA/BSc, 2 for MA/MSc/MBA, and 3 for DBA/PhD. Helfat and Martin (2015) argue that managers in brokerage positions have increased social capital, so we added the time spent on communication per day as a control, coded ordinally as 1 for no time, 2 for less than 30 min, 3 for 30–60 min, 4 for 1–2 h, and 5 for more than 2 h. These variables are intended to capture external communication and information-gathering.

3.3 Reliability and internal validity

To show the measurements’ reliability and validity, we calculated average variance extracted (AVE), maximum shared variance (MSV), maximum reliability (MaxR(H)), composite reliability (CR), and Cronbach’s alpha (α). All indicators show high quality of the constructs (Hair et al., 2010), as Table 1 shows. We calculated a series of model fit indicators, which confirm good model fit: χ2/df = 2.441 (χ2 = 246.519, df = 101), comparative fit index (CFI) 0.930, root mean square of approximation (RMSEA) 0.084, and standardised root mean square residual (SRMR) 0.0529 (Hu and Bentler, 1999).

To legitimise our structural equation modelling, we calculated convergent and discriminant validity statistics for all of the constructs in the model (Table 2). We again calculated a series of model fit indicators, which confirmed good model fit: χ2/df = 1.668 (χ2 = 1749.241, df = 1,049), CFI = 0.906, RMSEA = 0.057, and SRMR = 0.0653 (Hu and Bentler, 1999).

Table 3 shows the correlations among the multidimensional constructs, as well as the constructs’ means and standard deviations and the square roots of the AVEs. Following Fornell and Larcker (1981), we concluded that discriminant validity is given since the square roots of the AVEs were larger than the respective correlations of the constructs. Table 4 provides a detailed overview of the survey items for each of the constructs, as well as their factor loadings. To ensure discriminant validity, we ran the Heterotrait-monotrait (HTMT) ratio of correlation and the variance inflation factor (VIF) test (Kock, 2015). The HTMT ratios were well below the cut-off value of 0.85 that Kline (2023) suggests: HTMT_{sensing-seizing} = 0.672; HTMT_{sensing-reconfiguration} = 0.542; HTMT_{seizing-reconfiguration} = 0.626. We ran the VIF test first for first-order items (i.e. separate questions for each of the second-order constructs of sensing, seizing, and reconfiguration). The results indicated no multicollinearity, as the VIFs were well below the cut-off value of 10 (O’Brien, 2007), with the lowest value at 1.880 and the highest at 3.398. Then we ran the test for the aggregated second-order constructs of sensing, seizing, and reconfiguration, repeating the test three times for each of the constructs, and the highest value was 1.464. Consequently, we concluded that we faced no discriminant validity or multicollinearity issues. Table S2 in the supplementary materials shows the correlation table and the descriptive statistics for each item.

3.4 External validity

As part of the larger DC construct of capabilities (Adner and Helfat, 2003), DMCs must also be related to a firm’s performance (Sirmon and Hitt, 2009), so we regressed the DMC construct on firm performance to test its external validity. The test revealed that DMCs have a strong positive effect on firm performance (B = 0.935, β = 0.613, p < 0.001). At the same time, the model showed a good fit: χ2/df = 1.988 (χ2 = 681.994, df = 343); CFI = 0.923; RMSEA = 0.070; SRMR = 0.0705. We also ran a common method bias test that revealed no issues in this area. The results can be found in the supplementary file.

4. Results

We used IBM^® SPSS^® with a PROCESS plug-in by Hayes (2017) to test our complex mediation relationship. We ran four models: Model 1, the pure impact of positive affect on performance via sensing, seizing, and reconfiguration; Model 2, the same relationship as Model 1 but with control variables; Model 3, the pure impact of negative affect on performance via sensing, seizing, reconfiguration; and Model 4, the same relationship as Model 3 but with added control variables. Figure 2 shows the tested relationships with the respective constructs, Table 5 shows the direct and indirect effects, and Table S3, in the supplementary files, shows the control variables’ effects (with explanations.

As Table 5 shows, our hypotheses are supported: Both positive and negative affective states influence DMCs and, consequently, firm performance. Experience played a significant role in decreasing the sensing capacity in Model 1. The effects were also weakly significant in Model 3, as female managers showed lower seizing capacities than male managers and higher reconfiguration capacities (Table S3). Table S4 in supplementary materials summarises our hypothesis tests.

5. Discussion

Our findings contribute to the current discussion on DMCs (Helfat and Martin, 2015; Helfat and Peteraf, 2015; Martin, 2011) and affective states and “hot cognition” in managerial decision-making (Ashkanasy et al., 2017; Healey and Hodgkinson, 2017; Hodgkinson and Healey, 2011; Huy and Zott, 2019; Lebel, 2017). We show that managers who experienced recent negative affect have lower DMCs than those who experienced recent positive affect, thus offering two new aspects of this influence: that affective states may bias top managers toward changing their resource allocation and that DMCs may be the mechanism that explains how a manager’s current state is linked to firm performance. Here, we observe the chain reaction, as affect influences all three of the capacities that underlie a DMC, and sensing, whether limited or enhanced, influences seizing, reconfiguration, and, eventually, firm performance. Therefore, we can link managers’ affective state over their DMC to their firms’ performance.

Individuals are boundedly rational and have cognitive limitations, so they have to reduce environmental complexity to make sense of reality (Cyert and March, 1963). Emotions play an important role in this regard. As Damasio (2008, p. 127) observes,

Organisms whose brains only include those archaic structures and are devoid of evolutionarily modern ones–reptiles, for instance–operate such response selections without difficulty. One might conceptualise the response selections as an elementary form of decision making, provided it is clear that it is not an aware self but a set of neural circuits that is doing the deciding.

Netz et al. (2020) argue that the ability of individual managers in top and middle management teams to respond to unforeseen events affects communication and subsequent strategic actions. Our work shows how affect restricts or enhances the managerial abilities to sense opportunities and threats, seize an opportunity, and activate the firm’s resource base through a reconfiguration process.

For the past 15 years, the literature has measured DMCs using various proxies or qualitative data (Huy and Zott, 2019; Martin, 2011) or quantitative questionnaires (Åberg and Shen, 2020; Corrêa et al., 2019) that deviate from our vision of the formative operationalisation of the concept. For this reason, we proposed and tested an effective quantitative measure of and approach to the concept of DMC. We do not claim that these earlier approaches were wrong, but they are missing a quantitative survey with questions that relate to DMC at the manager level and that address Teece’s (2007) logic and the formative logic behind the construct. Our study provides a measure that fits these criteria and has a high degree of reliability and validity.

Wilden et al. (2013) treat the underlying capacities of sensing an opportunity or a threat, seizing it, and reconfiguring the asset base as non-interchangeable (formative) components that form the DC. Our research adds the input/output prism to this interrelationship and shows how a mediation mechanism within a DC works. Our mediation analysis reveals that the relationships among managerial sensing, seizing, and reconfiguration capacities are significant, while our discriminant validity test shows that these constructs are distinct, that they have a high degree of internal consistency, and that the scales can be used in separate investigations of each construct. Inspired by Wilden et al. (2013), we emphasise the necessity of a multidimensional formative approach not only to the construct of dynamic capabilities but also to DMCs.

6. Limitations

Our study presents two primary limitations. First, our respondents’ affective states could have influenced their evaluations of their capabilities. For instance, Sato and Kawahara (2011) show that individuals tend to bias their reports when they recall negative affective states, so our participants could have exaggerated their negative affective states’ impact on their firms’ performance. Nevertheless, as Spector et al. (2000, p. 79) argue figuratively in their study on negative affectivity and job stress, we “don’t throw out the baby with the bath water” but call for further longitudinal or experimental research to isolate this factor.

A second limitation lies in our having asked managers about their feelings in general, whereas asking specifically about how they felt regarding the actions of their firm would have introduced a different frame and is likely to have changed their responses (Kahneman, 2012). However, that approach could have added experimenter bias, as it introduces the themes of self-blame and other-blame (Smith and Lazarus, 1993) and could skew the PANAS scale by over- or under-representing items related to guilt, shame, and pride. In any case, our research was driven by the Affective Events Theory, which proposes that feelings in general accumulate and lead to changes in attitudes and actions, so an “in general” question about feelings was the best fit. Future research could try to address this issue.

7. Managerial implications

Our study has two lessons for management. First, affective states are not only a congruent part of every manager’s life but also a concept that helps to explain and, thus, improve the efficiency and effectiveness of a manager’s work. Equipped with the measurement tools from our paper, managers can evaluate their DMCs and the positive and negative affect they have experienced over a defined period. Reflecting on the state of their capabilities and affect might help them to incorporate “hot cognition” into their firms’ strategising processes.

Second, there are no good or bad emotions, but managers should acknowledge their expressions of various emotions and seek to direct them toward desirable outcomes. For instance, negative emotions might be helpful in the decision-implementation stage, as our findings suggest, whereas positive emotions tend to drive DMCs and, consequently, firm performance. Creating a positive emotional climate in a firm could become a priority of a good strategist.

8. Conclusion

In our research, we proposed a tool that can be used in future investigations of DMC and the relationships among its dimensions. Such investigations could benefit from using the tool to understand the micro-foundations of DMCs. For instance, Helfat and Peteraf (2015) propose several cognitive capabilities that are important for managerial sensing, seizing, and reconfiguration but leave unexamined whether they are moderated by affective states. Room for research on managerial sensing, seizing, and reconfiguration capacities remains: How does a transition between these capacities happen? What holds the three capacities together, and what can make them fall apart? Answers to these questions are to be found in future research. The present study only scratches the surface of affective states’ influence on DMCs, but our theory development and the measures we developed could help future researchers answer these questions.

Figures

Figure 1

Research model

Figure 2

Statistical tests model

Table 5

Direct and indirect effects in the model

		Model	1	PA		Model	2	PA(c)		Model	3	NA		Model	4	NA(c)
	Est.	p	LLCI	ULCI	Est.	p	LLCI	ULCI	Est.	p	LLCI	ULCI	Est.	p	LLCI	ULCI
Direct effects
a1 PA/NA → sensing	0.50	<0.001	0.39	0.62	0.50	<0.001	0.38	0.62	−0.21	0.002	−0.34	−0.08	−0.21	0.003	−0.34	−0.07
a2 PA/NA → seizing	0.35	<0.001	0.23	0.47	0.36	<0.001	0.23	0.48	−0.14	0.016	−0.25	−0.03	−0.15	0.010	−0.26	−0.04
a3 PA/NA → reconf-n	0.20	0.006	0.06	0.34	0.12	0.098	−0.02	0.26	0.14	0.018	0.02	0.25	0.11	0.064	−0.01	0.22
b1 sensing → performance	−0.10	0.182	−0.26	0.05	−0.12	0.136	−0.28	0.04	−0.09	0.256	−0.24	0.06	−0.09	0.226	−0.25	0.06
b2 seizing→performance	0.37	<0.001	0.21	0.53	0.37	<0.001	0.20	0.53	0.38	<0.001	0.23	0.54	0.38	<0.0001	0.22	0.55
b3 reconf-n → performance	0.221	0.003	0.08	0.37	0.22	0.005	0.07	0.37	0.28	<0.001	0.13	0.42	0.28	0.000	0.13	0.43
d21sensing → seizing	0.39	<0.001	0.26	0.51	0.35	<0.001	0.22	0.47	0.53	<0.001	0.42	0.65	0.49	<0.0001	0.38	0.61
d31sensing → reconf-n	0.22	0.003	0.07	0.36	0.19	0.009	0.05	0.33	0.29	<0.001	0.15	0.43	0.23	0.001	0.10	0.37
d32seizing → reconf-n	0.31	0.000	0.16	0.46	0.38	<0.001	0.23	0.52	0.42	<0.001	0.28	0.56	0.45	<0.0001	0.31	0.59
c’ (total direct effect)	0.16	0.033	0.01	0.31	0.18	0.021	0.27	0.33	−0.13	0.030	−0.25	−0.01	−0.14	0.028	−0.26	−0.02
Indirect effects
a1b1 → sensing	−0.05		−0.14	0.05	−0.06		−0.15	0.04	0.09		−0.02	0.09	0.02		−0.02	0.08
a1d21b2 → sensing and seizing	0.07		0.03	0.16	0.06		0.02	0.15	−0.04		−0.10	−0.01	−0.04		−0.09	−0.01
a1d31b3 → sensing and reconf-n	0.02		0.00	0.07	0.02		0.00	0.07	−0.02		−0.06	−0.00	−0.01		−0.04	−0.00
a2d32b3 → seizing and reconf-n	0.02		0.01	0.07	0.03		0.01	0.08	−0.02		−0.05	−0.00	−0.02		−0.05	−0.00
a1d21d32b3 → sensing and seizing and reconf-n	0.01		0.00	0.04	0.01		0.00	0.04	−0.01		−0.04	−0.00	−0.01		−0.04	−0.00
Total indirect effect	0.25		0.14	0.38	0.22		0.11	0.37	−0.09		−0.18	−0.01	−0.09		−0.18	−0.03
Total effect	0.42	<0.001	0.29	0.54	0.42	<0.001	0.29	0.55	−0.22	0.002	−0.35	−0.09	−0.23	0.001	−0.36	−0.09
R2, direct effects model	0.17				0.18				0.05				0.05
R2, total effects model	0.32				0.31				0.32				0.31
Delta R2	0.15				0.14				0.27				0.26

Note(s): PA stand for Positive Affect, NA for Negative Affect, (c) for a model with control variables, LLCI for the lower level for confidence interval, ULCI for the upper level for confidence interval, Est. for Estimates

Source(s): Table created by authors’

Table S1

Descriptive sample statistics

Measure	N	%	Measure	n	%
Firm size			Industry
SME (<250)	141	58.8	Construction	19	9.3
Big firm (≥250)	99	41.3	Education, research	15	7.4
Function			Energy and water supply	6	2.9
Management board	76	37.3	Financial and insurance services	18	8.8
R&D	10	4.9	Professional services (consulting, auditing, etc.)	23	11.3
Finance and controlling	13	6.4	Other commercial services	3	1.5
Production	16	7.8	Freelance, scientific, and technical activities	4	2.0
IT	46	22.5	Health care and welfare	13	6.4
Marketing and sales	13	6.4	Property and housing	6	2.9
Purchasing	3	1.5	Retail	12	5.9
Other	27	13.2	Hotel and hospitality	3	1.5
			IT, software, Internet	34	16.7
			Arts, entertainment, and recreation	3	1.5
			Logistics/transport (land, maritime, air)	5	2.5
			Media, publishing, and printing	3	1.5
			Broadcasting	1	0.5
			Telecommunications	2	1.0
			Production	17	8.3
			Other	17	8.3

Note(s): In the group “Other functions”, eight participants indicated that they were business owners. The rest of the answers related to other functions such as logistics (1), design (1), consultancy (1), system analysis (1), etc

Table S2

Correlation table for DMC with descriptive statistics

Item	M	SD	Sn01	Sn02	Sn03	Sn04	Sz01	Sz02	Sz03	Sz04	Sz05	Sz06	Rc01	Rc02	Rc03	Rc04	Rc05	Rc06
Sn01	5.966	1.221	–
Sn02	5.775	1.346	0.658	–
Sn03	6.132	1.243	0.707	0.587	–
Sn04	6.108	1.309	0.680	0.652	0.688	–
Sz01	6.167	0.968	0.376	0.256	0.366	0.347	–
Sz02	6.127	1.028	0.466	0.384	0.469	0.436	0.607	–
Sz03	6.069	1.029	0.382	0.331	0.366	0.397	0.671	0.634	–
Sz04	6.093	0.966	0.395	0.373	0.335	0.296	0.531	0.549	0.514	–
Sz05	6.098	0.998	0.516	0.413	0.490	0.467	0.539	0.598	0.608	0.538	–
Sz06	5.907	1.113	0.400	0.361	0.358	0.305	0.504	0.497	0.526	0.517	0.652	–
Rc01	5.495	1.384	0.409	0.454	0.285	0.416	0.269	0.350	0.408	0.371	0.293	0.308	–
Rc02	5.578	1.414	0.317	0.330	0.220	0.198	0.264	0.335	0.352	0.339	0.389	0.373	0.585	–
Rc03	5.554	1.358	0.323	0.403	0.251	0.437	0.323	0.362	0.466	0.385	0.320	0.334	0.726	0.576	–
Rc04	5.623	1.263	0.343	0.350	0.176	0.242	0.293	0.363	0.388	0.380	0.323	0.353	0.657	0.686	0.608	–
Rc05	5.574	1.346	0.455	0.466	0.364	0.320	0.251	0.406	0.373	0.360	0.457	0.408	0.658	0.682	0.620	0.658	–
Rc06	5.691	1.274	0.452	0.488	0.390	0.372	0.341	0.376	0.444	0.368	0.377	0.383	0.590	0.589	0.663	0.591	0.716	–

Note(s): Sn stands for sensing, Sz for seizing, and RC for reconfiguration

Table S3

Control variables effects

			Model 2	PA(c)					Model 4	NA(c)
	Sensing		Seizing		Reconfiguration		Sensing		Seizing		Reconfiguration
	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p
Control variables
Gender	−0.03	0.803	−0.38	<0.001	0.19	0.089	−0.03	0.806	−0.37	0.001	0.20	0.079
Firm size	0.00	0.631	0.00	0.693	0.00	0.229	0.00	0.223	0.00	0.384	0.00	0.197
Age	0.01	0.368	0.01	0.127	−0.03	<0.001	−0.01	0.234	−0.00	0.857	−0.09	<0.0001
Experience	−0.02	0.011	−0.00	0.490	−0.01	0.235	−0.01	0.066	0.00	0.905	−0.01	0.193
Education	−0.08	0.410	0.09	0.291	−0.09	0.840	−0.03	0.800	0.13	0.140	−0.02	0.798
Brokerage	0.08	0.120	0.06	0.185	0.02	0.681	0.15	0.007	0.09	0.063	0.02	0.633

Note(s): PA stand for Positive Affect, NA for Negative Affect, (c) for a model with control variables, Est. for Estimates

Table S4

Hypotheses tests

Hypotheses #	Relationship	Result
H1a	PA → sensing	Supported
H1b	PA → seizing	Supported
H1c	PA → reconfiguration	Supported
H1d	PA ⇢ DMC ⇢ performance	Supported
H2a	NA → sensing	Supported
H2b	NA → seizing	Supported
H2c	NA → reconfiguration	Supported
H2d	NA ⇢ DMC ⇢ performance	Supported

Notes

1. In contrast to scholars who follow Teece (2007) in operationalising DMC as owned by a collective entity like a firm’s board (Åberg and Shen, 2020), we propose a measurement that addresses a single manager’s DCs.

Supplementary material

Common method bias

Since we collected both the dependent and the independent variables from the same data source, we tested for common method bias using various test methods. We constructed a model including both the DMCs construct and the PANAS. First, we ran Harman’s single-factor test (Podsakoff and Organ, 1986). This factor explains 30.82% of the variance, which is far below the cut-off value of 50%. Nevertheless, since this test provides no statistical control for method effects, we conducted a common latent factor test (Richardson et al., 2009). The variance explained by a common latent factor is below 2%. The introduction of the latent factor had no significant influence on the model fit: χ²/df = 1.700 (χ² = 1786.911, df = 1,051); CFI = 0.902; RMSEA = 0.059; SRMR = 0.0534.

Table 1

Reliability and validity of the construct dynamic managerial capabilities

Construct	α	CR	AVE	MSV	MaxR(H)
Dynamic managerial capabilities (Second order)	0.77	0.77	0.53	–	0.78
Managerial sensing	0.89	0.89	0.66	0.41	0.89
Managerial seizing	0.89	0.89	0.57	0.41	0.94
Managerial reconfiguration	0.91	0.91	0.64	0.35	0.96

Source(s): Table created by authors’

Table 2

Reliability and validity of the constructs used in the research model

Construct	α	CR	AVE	MSV	MaxR(H)
Managerial sensing	0.89	0.89	0.66	0.41	0.95
Managerial seizing	0.89	0.89	0.57	0.41	0.96
Managerial reconfiguration	0.91	0.91	0.64	0.35	0.97
Performance†	0.89	0.93	0.81	0.32	0.99
Positive affect	0.90	0.90	0.47	0.39	0.91
Negative affect	0.95	0.95	0.66	−0.08	0.99

Note(s): ^†indicates a second-order construct

Source(s): Table created by authors’

Table 3

Correlations and average factor loadings

Construct	(1)	(2)	(3)	(4)	(5)	(6)
(1) Managerial sensing	0.81‡
(2) Managerial seizing	0.64	0.75‡
(3) Managerial reconfiguration	0.54	0.59	0.80‡
(4) Performance†	0.33	0.57	0.48	0.90‡
(5) Positive affect	0.57	0.63	0.52	0.46	0.69‡
(6) Negative affect	−0.24	−0.29	−0.03	−0.24	−0.17	0.81‡

Note(s): ^†indicates a second-order construct; ^‡ indicates average factor loading

Source(s): Table created by authors’

As we can see, experience played a significant role in the model with a positive affect – experience decreased the sensing capacity. This goes in line with path dependency theory (David, 1985; Sydow et al., 2009). The effects were also weakly significant in the model with the negative affect. Interestingly, the experience did not impact the seizing and reconfiguration capacities. Age had in both models a negative effect on reconfiguration. We did not find an explanation for this observation.

Another interesting finding is the impact of gender on seizing and reconfiguration capacities. Female managers showed lower seizing capacities and higher reconfiguration capacities. This research contributes to the seminal research of Bogodistov et al. (2017) who showed that female managers weakened the sensing capacity but enhanced it when leading gender-diverse teams. Our study showed the effects of gender on the seizing and reconfiguration capacities of DMC.

Table 4

Dynamic managerial capabilities: items and factor loadings

Construct	Item	Loading (ML)	Loading (PCA)
Dynamic managerial capability†	In my company …
Managerial sensing	I can fully understand the impact of the internal and external environment	0.85	0.88
	I can feel (intuitively perceive) major potential opportunities and threats	0.75	0.83
	I have good observation and judgement ability	0.82	0.86
	My judgement and observation ability are demanded in our organization	0.81	0.88
	In my company …
Managerial seizing	I invest time and effort in finding solutions for our customers	0.74	0.81
	I invest time and effort in finding solutions for our organization	0.78	0.81
	I invest time and effort in finding solutions for my recipients	0.78	0.83
	I adopt the best practices in management	0.69	0.76
	I adopt the best practices in the field which is most relevant to my work	0.77	0.82
	I change our practices when customers or our management feedback gives us a reason to change	0.70	0.77
	How often did you personally carry out the following activities last year?
Managerial reconfiguration	Implementation of new kinds of management methods	0.81	0.84
	New or substantially changed marketing method or strategy	0.77	0.82
	New or substantial changes to working methods or to our unit’s strategy and tactics	0.79	0.84
	Substantial renewal of business processes	0.79	0.84
	Substantial renewal of working routines and processes	0.83	0.87
	New or substantially changed ways of achieving our targets and objectives	0.79	0.83

Note(s): ^† – all sub-constructs were approached at the individual level; ML = maximum likelihood; PCA = principal component analysis

Source(s): Table created by authors’