To exploit or explore research on the relationship between performance gap and ambidexterity innovation

Bạn đang xem bản rút gọn của tài liệu. Xem và tải ngay bản đầy đủ của tài liệu tại đây (1.51 MB, 15 trang )

Trang 1<div class="page_container" data-page="1">

Full Terms & Conditions of access and use can be found at

Technology Analysis & Strategic Management

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/ctas20

To exploit or explore? Research on the relationshipbetween performance gap and ambidexterity

Qunpeng Fan, Songsong Cheng, Dongphil Chun & Yan Zhang

To cite this article: Qunpeng Fan, Songsong Cheng, Dongphil Chun & Yan Zhang (21

Feb 2024): To exploit or explore? Research on the relationship between performance gap and ambidexterity innovation, Technology Analysis & Strategic Management, DOI: 10.1080/09537325.2024.2319596

To link to this article: online: 21 Feb 2024.

Submit your article to this journal

Article views: 29

View related articles

View Crossmark data

</div>Trang 2<div class="page_container" data-page="2">

To exploit or explore? Research on the relationship between performance gap and ambidexterity innovation

Qunpeng Fana,b, Songsong Chengc, Dongphil Chunband Yan Zhangd

School of Economics, Management and Law, Jilin Normal University, Siping, People’s Republic of China;b

Graduate School of Management of Technology, Pukyong National University, Busan, Korea;cSchool of Economics and Finance, South China University of Technology, Guangzhou, People’s Republic of China;d

School of Management, Guangzhou College of Technology and Business, Guangzhou, People’s Republic of China

What type of innovation strategy a ﬁrm will adopt when facing the performance gap, and how this strategy will aﬀect the ﬁrm’s long-term growth, are important theoretical questions. Based on the behavioural theory of theﬁrm, we aim to investigate the mechanism through which the performance gap inﬂuences ambidexterity innovation from an industrial aspiration perspective. Utilising a sample of A-share manufacturing listedﬁrms in China from 2007 to 2018, we employ Stata 16.0 to test hypotheses. The ﬁndings reveal an inverted U-shaped relationship between performance gap and exploitative innovation and a U-shaped relationship between performance gap and exploratory innovation. That is, when ﬁrms experience a gap between their performance and industrial aspiration, they tend to prioritise exploitation over exploration as their innovative approach. Exploitative innovation and exploratory innovation have an inverted U-shaped and U-shaped impact on the value recreation capability, respectively, under the performance gap. Based on the exploitation-exploration ambidexterity innovation framework, we empirically validate the ‘local-beyond local’ problem search logic posited by the behavioural theory of the ﬁrm. Furthermore, by comparing the diﬀerentiated recovery eﬀects of exploitative and exploratory innovations, we address the sequential allocation challenges in innovation activities for

In the aftermath of the emergence of unforeseen‘black swan’ and ‘gray rhino’ events, and the sub-sequent profound repercussions, numerous enterprisesﬁnd themselves ensnared in the quagmire of diminishing market share and trailing behind benchmark companies in terms of performance. Scho-lars characterise this adverse scenario as a performance gap. According to the behavioural theory of theﬁrm (BTOF), when organisational performance fails to meet predeﬁned aspiration, enterprises embark on a quest to identify problems and strategically address performance shortcomings (Cyert and March1963). Consequently, considerable scholarly attention has been devoted to exam-ining the problem-searching behaviour and strategic change patterns of underperformingﬁrms. Among these, innovation, a classic strategic behaviour employed byﬁrms to establish a competitive advantage (Le Roy, Robert, and Hamouti 2022; Morandi Stagni, Fosfuri, and Santaló 2021), has

CONTACTYan of Management, Guangzhou College of Technology andBusiness, Guangzhou 510850, People’s Republic of China

class="text_page_counter">Trang 3<div class="page_container" data-page="3">

garnered widespread attention and undergone extensive scrutiny. However, existing research on the relationship between performance gap and innovation yields inconclusiveﬁndings.

To address the inconsistency in existingﬁndings, we ﬁrst undertake a re-evaluation of the aspira-tion reference point, aiming to redeﬁne the characterisation of the performance gap by substituting the mixed model with a separated model. Previous studies have commonly depicted the perform-ance gap by ascribing varying weights to historical aspiration and industrial aspiration, favouring higher weights for the latter and lower weights for the former to formulate a mixed model (Parker, Halgin, and Borgatti2016). However, it is essential to recognise that historical aspiration centre on the organisation’s own past performance, while industrial aspiration revolves around the performance of the organisation’s competitors (Kim, Finkelstein, and Haleblian2015). These dis-tinct dimensions carry diﬀerent managerial implications (Manzaneque et al.2020; Mount and Baer

2022). Building on this insight, we disentangle industrial aspiration from historical aspiration and adopt a separated model represented by the industrial aspiration to characterise the performance gap. Second, we serve to elucidate the theoretical foundation and employ the problem search logic intrinsic to BTOF. The pivotal concept of BTOF is the ‘local-beyond local’ problem search logic. According to Cyert and March (1963), when a performance gap arises, enterprises typically engage in problem searches within the local realm of the organisation. Only if the incremental sol-ution of updating the existing product proves insuﬃcient to resolve the performance gap, will the enterprise go beyond its local expertise to explore newer and broader technologies. Building on the logic of problem search, weﬁnd that the innovation activities of underperforming ﬁrms are complex and diverse, aligning with the connotation of ambidexterity innovation. Hence, we introduce the third research extension, wherein, grounded in March’s ambidexterity theory, we classify innovation into exploitative innovation (ELI) and exploratory innovation (ERI) (Jansen, Van Den Bosch, and Vol-berda2006). ELI involves the reconstruction of local knowledge and optimisation of existing product technology, ERI strives to beyond local knowledge boundaries and leverage novel insights (Donbe-suur et al.2023). Building upon the aforementioned analysis, we employ the problem search logic to scrutinise the strategic decision-making behaviour associated with ambidexterity innovation in the context of the industrial performance gap, with industrial aspiration serving as the pivotal reference point.

Furthermore, the complete problem search chain encompasses the performance gap, organis-ational response, and value recreation. The achievement of value recreation in ﬁrms grappling with the performance gap is a topic of considerable interest to enterprises. Value recreation involves the operational recovery or enhancement of proﬁtability for an enterprise following a downturn (Santana, Valle, and Galan 2017). While the performance gap inﬂuences enterprises’ inclination towards ELI and ERI, it remains uncertain whether ambidexterity innovation can enhance the organ-isation’s value recreation capacity. Therefore, aligning with the primary analytical thread of problem search, we not only delve into the ambidexterity innovation strategies of underperformingﬁrms but also scrutinise the impacts of diﬀerent innovation behaviours on value recreation capacity. Speciﬁ-cally, we analyse whether ELI and ERI exhibit diﬀerential eﬀects on an organisation’s value recreation capacity in the aftermath of the performance gap.

The paper’s research contributions are reﬂected in three aspects: First, a separated model is employed to distinguish industrial aspiration from historical aspiration, with a speciﬁc emphasis on the relationship between the industrial performance gap andﬁrm innovation. The study contrib-utes to the formation of consistent conclusions in performance feedback-related studies. Second, we incorporate the exploitation-exploration ambidexterity innovation framework into the performance gap model. This not only conﬁrms the ‘local-beyond local’ problem search logic of BTOF but also veriﬁes the complexity and diversity of enterprises’ innovative response mechanism to dilemmas. Third, exploring the distinct value recreation eﬀect of ambidexterity innovation within enterprises facing the performance gap, not only guides the resolution of sequential allocation challenges in innovation activities for underperforming enterprises but also addresses the missing logic in the latter part of the‘performance gap-organizational response-value recreation’ framework.

</div>Trang 4<div class="page_container" data-page="4">

2. Literature and hypothesis 2.1. Performance gap

Performance gap emerges through a comparison of the enterprise’s present performance against its historical benchmarks (historical aspiration) and/or the performance of its industrial peers (industrial aspiration). As research progresses, scholars suggest focusing on separated model rather than a mixed model, especially the industrial performance gap (Yu, Minniti, and Nason 2019). This rec-ommendation stems from scholarly observations indicating that, owing to the pressures of iso-morphism, the industrial performance gap exerts a more signiﬁcant inﬂuence on ﬁrms’ subsequent strategic responses than the historical performance gap does. Dong’s (2021) simulation results further aﬃrm that, in most cases, referencing industrial aspiration optimally informs ﬁrms’ technology-related strategic choices. Thus, we adopt the industrial aspiration perspective and deﬁne the performance gap as a state of discrepancy wherein a ﬁrm’s actual performance falls below its degree of industrial aspiration.

2.2. Ambidexterity innovation

Ambidexterity innovation refers toﬁrms participating in both exploitative and explorative activities (Jansen, Van Den Bosch, and Volberda2006). ELI aims at extending current technology and seeking greater eﬃciency and improvements to improve existing product-market standings. ERI aims to identify new market trends and engage in innovation beyond the existing product-market domain. The extant view of ambidexterity innovation recognises that exploitation and exploration, two contradictory yet interrelated activities, can separate and coexist over time (Andriopoulos and Lewis2009; Lennerts, Schulze, and Tomczak2020; Osiyevskyy, Shirokova, and Ritala2020). The con-ceptualisation of separation builds on the competition perspective and emphasises that exploitation and exploration usually compete for organisational resources. The conceptualisation of coexistence proposes that exploitation and exploration are considered interdependent activities.

2.3. Problem-based search logic

Problem-based search constitutes a core concept in BTOF. Jung and Lee (2016) posit that aﬁrm’s local search entails seeking knowledge akin or closely related to the existing expertise of theﬁrm. Organisations also tend to learn more eﬀectively in local areas of familiar or related technologies (Posen et al.2018). It is only when a satisfactory solution proves elusive through local search that ﬁrms ﬁnd the motivation to transcend their local inclinations and venture beyond their current narrowﬁeld of expertise in problem search. In contrast to local search, beyond local search delves into technologies that are unfamiliar, irrelevant, and distant from the ﬁrm’s existing knowledge base. This exploration into uncharted territory is more likely to result in breakthrough solutions for theﬁrm (Gambeta, Koka, and Hoskisson2019).

2.4. Performance gap and ambidexterity innovation

Based on the problem search logic of BTOF, we propose the existence of an inverted U-shaped relationship between the performance gap and ELI. Speciﬁcally, when organisational performance falls below the industrial aspiration and still in the neighbourhood of the industrial aspiration level,ﬁrms are compelled to seek incremental solutions to address competitive deﬁciencies in the current market (Ref and Shapira2017). This approach opens up opportunities for value recreation and the attainment of desired goals. When organisational performance falls below industrial aspira-tion but well below industrial aspiraaspira-tion level, the organisaaspira-tion must transcend localised search eﬀorts to ﬁnd eﬀective problem-solving strategies (Diwei Lv et al. 2021). A severe performance

</div>Trang 5<div class="page_container" data-page="5">

gap undermines the eﬃcacy of business management and strategy execution (Cheng et al.2022). While maintaining the current product model can improve operational eﬃciency, it fails to address the underlying business predicament and may lead to increased external business pressures. On the other hand, adopting unconventional product practices and breaking away from established technology paradigms has been shown to enhance performance. Psychological research has further shown that the accumulation of failures weakens the self-eﬃcacy of individuals with limited ration-ality. Consequently, poor business performance diminishes decision-makers’ conﬁdence in navigat-ing uncertainty usnavigat-ing existnavigat-ing organisational strategies (Tarakci et al.2018), leading them to favour deviations from the original technological trajectory in pursuit of innovative activities. In summary, as the performance gap widens,ﬁrms tend to adopt a ‘local-beyond local’ logic of problem search, and the phenomenon of ELI appears to peak and subsequently decline. Thus, we predict the following:

H1: There is an inverted U-shaped relationship between performance gap and ELI.

Based on the simplicity and process characteristics of problem-based search, our argument suggests that the performance gap leads to a U-shaped change inﬁrms’ ERI, wherein ERI ﬁrst decreases and then increases. As March (1991) emphasises, exploration necessitates elements such as‘search, vari-ation, experimentvari-ation, play,ﬂexibility, discovery, and innovation’ (71). Thus, ERL involves breaking free from organisational strategic constraints, deviating from established technological practices, and transcending local organisational boundaries (Lei, Khamkhoutlavong, and Le2021). When the performance falls below industrial aspiration and still in the neighbourhood of industrial aspiration level, decision-makers tend to focus their attention on localised aspects of the organisation. The narrow scope of technological search in this situation is inconsistent with the innovative foundation of experimentation, creativity, and risk-taking, thereby weakening ERI. When organisational perform-ance falls below industrial aspiration but well below industrial aspiration level, decision-makers are more motivated to explore problems beyond the local realm. Engaging in cross-boundary activities across industries and technologies can assistﬁrms in creating new cognitive schemas for technology. This, in turn, stimulatesﬁrms to break away from conventional technology practices and challenge the established technological architecture, eﬀectively fostering ERI. In summary, as the performance gap increases, ERI exhibits a U-shaped curve change. Thus, we predict the following:

H2: There is a U-shaped relationship between performance gap and ERI.

3. Research methods 3.1. Data source

The paper uses data from the CSMAR database and focuses on the listed manufacturing businesses in Shanghai and Shenzhen A-shares from 2007 to 2018 as the research sample. The choice of the manufacturing industry is driven by two reasons. First, it helps to reduce the biased impact of indus-try diﬀerences on organisational innovation decisions. Second, the manufacturing industry is a subject of innovation subjects in China, making it a suitable area for this study. During the initial sample screening process, we exclude the samples of ST and * STﬁrms, as well as the ﬁrms with missing critical data. After this screening process, theﬁnal dataset comprises 8857 valid samples. To ensure data quality, all continuous variables have been winsorised, meaning that extreme values were adjusted to the 1% and 99% quantiles.

3.2. Measurement 3.2.1. Performance gap

We use Return on Assets (ROA) as a measure of ﬁrm performance. Drawing on Chen’s (2008) research, we employ two procedures to measure the performance gap. First, we calculate the

</div>Trang 6<div class="page_container" data-page="6">

industrial aspiration level (IA) using Formula (1). This involves computing the weighted average value of theﬁrm’s industry median level in the t-2 (weight of 0.4) and t-1 period (weight of 0.6).

1)IAt represents industrial aspiration level in period t, IPert-1 stands for industry performance median in period (t-1), and IPert-2stands for industry performance median in period (t-2).

Second, we assess the degree of the performance gap by subtracting the IA from the actual per-formance. If the result is less than 0, it indicates that theﬁrm is in the performance gap; if it is greater than or equal to 0, it means that theﬁrm is performing above par. To handle the data, we use the truncated dummy variable method in Formula (2) (Ref and Shapira2017; Tarakci et al.2018). Pert represents the actual performance in period t, and I is the dummy variable constructed for this study. If the result of (Pert- IAt) is less than 0, I take the value 1. If the result is greater than or equal to 0, I take the value 0. Finally, we take the absolute value of PGtto capture its magnitude.

PGt= I (Pert- IAt) (2)

3.2.2. Ambidexterity innovation

Based on the Accounting Standards for Business Enterprises in China, Bi et al. (2017) devised a method to measure ambidexterity innovation. In this method, investment in the research stage is considered ERI investment, while investment in the development stage is regarded as ELI invest-ment. Consistent with their approach, we measure ELI and ERI using capitalisation and expense expenditure of R&D investment.

3.2.3. Control variables

We set up several control variables: the integration of the chairman and CEO (Duality), political con-nections (PC),ﬁrm age (Age), ﬁrm size (Size), the proportion of independent directors (Ind), TMT shareholding ratio (TMTshare), equity concentration (EC), slack resources (Slack), the nature of prop-erty right (NPR), the threat of bankruptcy (Bkrupt), market environmental dynamics (Market). We also set up an annual dummy variable to account for the eﬀect of year evolution on ﬁrm innovation decisions.

3.3. Models

Based on the research hypothesis, we set the following panel data models:

Y= b0+ b1PG+ b2PG2+ b3CV+1 (3) Where Y is the dependent variable, represented by ELI and ERI; PG stands for performance gap, PG2is the quadratic term of the performance gap, CV is a set of control variables, and⍰ is the residual term.

4. Results

4.1. Descriptive statistics

Table 1shows descriptive statistics and correlations for the main variables. The average value of ELI is 0.018, with a standard deviation of 0.064, indicating that the samples exhibit variability in ELI.

Table 1.Results of descriptive statistics and correlation analysis.

</div>Trang 7<div class="page_container" data-page="7">

Similarly, the average value of ERI is 0.084, with a standard deviation of 0.070, highlighting signiﬁcant diﬀerences in ERI among the samples. The average performance gap is 0.017, indicating that, on average, there is a 1.7% diﬀerence between the ﬁrm’s performance and its industrial aspiration in the study. Moreover, all coeﬃcients are less than 0.5, indicating that the paper does not encounter serious multi-collinearity issues.

4.2. Regression analysis

To mitigate the potential bias caused by missing variables, we adopt theﬁxed eﬀect model for regression analysis (Wang et al.2021; Yu, Minniti, and Nason2019).Table 2presents the results of the regression analysis.

Model 1 includes only the control variables. Model 2 and Model 3 incorporate the independent variable performance gap and the quadratic term of the performance gap, respectively. The regression coeﬃcient between the performance gap and ELI is signiﬁcantly negative at the 1% level (β = −.098, p < 0.01), indicating there is a negative linear relationship between performance and ELI. And, the regression coeﬃcient between the quadratic term of the performance gap and ELI is signiﬁcantly negative at the 5% level (β = −1.345, p < 0.05). This suggests a signiﬁcant inverted U-shaped relationship between the performance gap and ELI.

ERI is used as the dependent variable in Models 4–6. The analysis reveals that the performance gap has a signiﬁcant positive impact on ERI (β = .315, p < 0.01), indicating a signiﬁcant positive linear relationship between the performance gap and ERI. The regression coeﬃcient between the quadratic term of the performance gap and ERI is signiﬁcantly positive at the 1% level (β = 1.829, p < 0.01). This suggests a signiﬁcant U-shaped relationship between the performance gap and ERI. Based on the regression results, we adopt the curve test developed by Lind and Mehlum (2010) to verify the existence of the inverted-U and U-shaped relationships. The test involves two steps: First, data analysis was used to calculate the slopes of the low-level performance gap curve and the level performance gap curve. If the slope is positive on the low-level curve and negative on the high-level curve, it indicates a partially established inverted-U shaped relationship. Conversely, if the slope

Table 2.Test of the relationship between performance gap and ELI/ERI.

</div>Trang 8<div class="page_container" data-page="8">

is negative on the low-level curve and positive on the high-level curve, it indicates a partially estab-lished U-shaped relationship. Second, the Fieller and Delta conﬁdence interval test was performed to examine whether the extremes (poles) of the curve fall within the high and low-value intervals of the performance gap. If the poles fall within these intervals, it conﬁrms the existence of the curve relationship. For the relationship between the performance gap and ELI, the slope on the curve of the low-level performance gap is positive (0.073), while the slope on the curve of the high-level per-formance gap is negative (−0.388). The value of the pole is 0.027, which falls within the interval of [0, 0.171]. This indicates that the inverted U-shaped relationship between the performance gap and ELI indeed exists, thus conﬁrming hypothesis 1. Regarding the relationship between the performance gap and ERI, the slope on the curve of the low-level performance gap is negative (−0.032), and the slope on the curve of the high-level performance gap is positive (0.617). The value of the extreme point is 0.009, which also falls within the interval of [0, 0.171]. This conﬁrms the existence of the U-shaped relationship between the performance gap and ERI. Therefore, hypothesis 2 was tested and supported.

To quantify the strength of the correlation, we use the eﬀect size, speciﬁcally Cohen’s (f 2), as the evaluation indicator. In this paper, we apply the eﬀect size to assess the performance gap and its quadratic term on both ELI and ERI. After performing the calculation, weﬁnd that in the context of the performance gap and ELI, Cohen’s (f 2) for the linear relationship is 0.108, while Cohen’s (f 2) for the nonlinear relationship is 0.396. This indicates that the nonlinear impact of the performance gap on ELI is greater than its linear impact. Similarly, concerning the performance gap and ERI, Cohen’s (f 2) for the linear relationship is 0.292, whereas Cohen’s (f 2) for the nonlinear relationship is 0.761. These results suggest that the performance gap has a stronger U-shaped impact on ERI than its linear impact. Thus, the results strongly support both H1 and H2.

4.3. Robustness test

4.3.1. Changing the measurement of an independent variable

To minimise the potential impact of weight assignment on the results, we recalculate the industrial aspiration and performance gap using the weight values of 0.7 and 0.3. The model 1–2 inTable 3

presents the regression results. In model 1, the regression coeﬃcient of the quadratic term of the performance gap is not signiﬁcant (p > 0.1). Therefore, hypothesis 1 is not further supported. In model 2, the quadratic term of the performance gap is signiﬁcantly correlated with the regression coeﬃcient of ERI (β = .068, P < 0.1). This result demonstrates that the performance gap and ERI exhibit a U-shaped relationship. Thus, H2 is supported again.

4.3.2. Changing the regression model

We adopt the Tobit model to test the relationship between performance gap and ambidexterity innovation. This choice is based on the data structures of ELI and ERI, which exhibit a truncated

Table 3.Robustness test.

Changing the measurement of independent

</div>Trang 9<div class="page_container" data-page="9">

feature with 0 as the lower limit and 1 as the upper limit. The model 3–4 inTable 3shows the empiri-cal results. Despite the model change, theﬁnding still reveals a signiﬁcant inverted U-shape relation-ship between the performance gap and ELI (β = −1.062, P < 0.05), as well as a signiﬁcant U-shaped relationship between the performance gap and ERI (β = 2.740, P < 0.01). These results further support H1 and H2.

4.4. Endogeneity testing

4.4.1. Lagging the independent variables for one period

To address concerns about reverse causation, we introduce a lagged performance gap variable. The empirical results are shown inTable 4. In model 1, the quadratic term of the lagged performance gap is signiﬁcantly and negatively correlated with ELI (β = -.688, P < 0.1), demonstrating that H1 is still supported. In model 2, the quadratic term of the lagged performance gap is signiﬁcantly and posi-tively correlated with ERI (β = .883, P < 0.1), which supports H2 once more.

4.4.2. Propensity scores matching method (PSM)

To mitigate the potential biases arising from sample selection and endogeneity issues, we employ the PSM method to further examine the inﬂuence of the performance gap on ELI and ERI. The approach involves dividing the sample into a treatment group (high-performance gap) and a control group (low-performance gap) based on the median performance gap threshold and using the control variables in the model (3) as matching variables. Subsequently, nearest-neighbor match-ing is employed to match samples accordmatch-ing to their propensity score values. Finally, the matched samples are utilised for re-regression analysis of the model (3). Models 3 and 4 inTable 4 demon-strate that even after employing these techniques, the coeﬃcients of the squared term of perform-ance gap remain signiﬁcantly negative (β = −1.161, P < 0.1) for ELI and positive (β = 1.890, P < 0.05) for ERI, which aligns with our earlier regressionﬁndings.

5. Further analysis

Hoskisson et al. (2017) have highlighted the signiﬁcance of examining whether a ﬁrm’s strategic behaviour creates value in the context of the performance gap. As such, we aim to conduct a com-prehensive theoretical analysis and empirical investigation to explore the impact of ELI and ERI on theﬁrm’s value recreation capability.

The capability ofﬁrms to enhance value recreation within the performance gap zone, as they approach industrial aspirations, hinges on rapid recovery. This entails the organisation’s capability to swiftly address consumer demand through localised search and product upgrades. By intensifying investment in ELI and fostering greater iteration of established technology,ﬁrms not only uncover fresh market opportunities but also crucially cultivate a competitive advantage. However, the

Table 4.Endogeneity testing.

Lagging the independent variables for one

</div>Trang 10<div class="page_container" data-page="10">

introduction of ERI might disrupt the continuity and integrity of the organisation’s established tech-nology, potentially delaying market capture and rebound. In the performance gap zone further away from industrial aspirations, theﬁrms’ capacity to bolster value recreation relies on full recovery. In this scenario, organisations must transcend localised search and systematically develop new pro-ducts to establish sustainable core competencies. Overreliance on ELI may leadﬁrms into an iterative trap, perpetually cycling through established technologies without substantial progress. Instead, to facilitate organisational leapfrogging into new technology tracks,ﬁrms must augment investment in ERI. This strategic move not only creates new internal and external investment opportunities but also serves as a pivotal factor in constructing dynamic capabilities and orchestrating a remarkable resur-gence from adversity.

To assess the impact of ELI and ERI on closing the performance gap, we employ the Jensen Index as a robust measure of theﬁrm’s value recreation capacity (Morrow Jr et al.2007). The index con-siders both ordinary and logarithmic rates of return and oﬀers a potential perspective on enterprises to obtain high expected returns.Table 5reports the results with the panel eﬀect model. The ﬁndings show an inverted U-shaped relationship between ELI and value recreation capacity under the per-formance gap (Figure 1), as well as a U-shaped relationship between ERL and value recreation capacity (Figure 2). TheFigure 1and 2 are based on Models 1 and 3 ofTable 5, respectively.

6. Discussion and conclusion

Based on the data of listed companies in China, we explore the allocation patterns ofﬁrms between ELI and ERI concerning the performance gap. Additionally, we analyse the roles of ELI and ERI in enhancing the value recreation capabilities ofﬁrms. The research ﬁndings are as follows: (1) There exists an inverted shaped relationship between the performance gap and ELI, as well as a U-shaped relationship between the performance gap and ERI. (2) In the context of the performance gap, there is an inverted U-shaped relationship between ELI andﬁrms’ value recreation capability and a U-shaped relationship between ERI and value recreation capability.

6.1. Theoretical contributions

First, we employ a separated model to analyse the mechanism through which the performance gap, relative to industrial competitors, inﬂuences ﬁrms’ ambidexterity innovation. It contributes to resol-ving inconsistent conclusions in performance gap research. We thoroughly dissect the performance gap, employing a separated model that dissects industrial aspiration from historical aspiration. By doing so, it explores the innovation eﬀect of the industrial performance gap, a facet that has been accorded signiﬁcant weight by researchers. The ﬁndings suggest that underperforming peer

Table 5.Impact of performance gap on value recreation capability.

Jensen Index calculatedbased on ordinary return

Jensen Index calculatedbased on logarithmic

return rate

Jensen Index calculatedbased on ordinary return

Jensen Index calculatedbased on logarithmic return

Notes: Figures in brackets are T value, *p < 0.1, **p < 0.05, ***p < 0.01.

</div>