Professional Documents
Culture Documents
Kump Et Al. Toward A Dynamic Capabilities Scale - Measuring
Kump Et Al. Toward A Dynamic Capabilities Scale - Measuring
doi: 10.1093/icc/dty054
Original Article
Abstract
To date, no standard scale exists for measuring dynamic capabilities. This limits the comparability of
empirical findings and impairs data-based theory development. This article presents the development
of a 14-item scale based on Teece’s (2007, Strategic Management Journal, 28, 1319–1350) well-
established dynamic capability framework, assessing sensing, seizing, and transforming capacities. It
describes the rigorous empirical scale development procedure comprising the steps of (i) item gener-
ation, (ii) scale purification (n ¼ 269), and (iii) scale confirmation (n ¼ 307). The scale shows high reli-
ability and validity and is a solid predictor of business and innovation performance.
JEL classifications: O31, C83
1. Introduction
Since its introduction approximately two decades ago (Helfat, 1997; Teece et al., 1997; Eisenhardt and Martin,
2000), the concept of dynamic capabilities (DC) has received great attention from management scholars (for reviews
see, Ambrosini and Bowman, 2009; Barreto, 2010; Di Stefano et al., 2010; Vogel and Güttel, 2013; Wilden et al.,
2016; Schilke et al., 2018). Defined as organizational capabilities that allow firms to “build and renew resources and
assets [. . .], reconfiguring them as needed to innovate and respond to (or bring about) changes in the market and in
the business environment” (Teece, 2014: 332), DC are crucial for firms to achieve strategic change and renewal
(Helfat et al., 2007; Agarwal and Helfat, 2009).
A rigorous literature review by Schilke et al. (2018) revealed two important observations: first, in contrast with
earlier days when most DC studies were conceptual, the majority of DC research is now empirical; second, while
C The Author 2018. Published by Oxford University Press on behalf of Associazione ICC. All rights reserved.
V
2 Toward a dynamic capabilities scale
recently researchers (Stadler et al., 2013; Girod and Whittington, 2017) have started to explore the potential of
employing proxies for measuring DC, still 33% of all publications on DC they reviewed report findings from survey
capabilities, DC may contribute to competitive advantage, but they are not sufficient for sustained firm performance
(Helfat et al., 2007; Wilden et al., 2016; Schilke et al., 2018).
innovation). These higher-order capabilities may enable firms to deploy and modify lower-order DC in the form of
simple rules. As a second circumstance under which the approaches are in alignment also in high-velocity markets,
selection from a variety of strategic options. Seizing capacity within an organization is high if the organization is able
to decide whether some information is of potential value, to transform valuable information into concrete business
Note: The scale purification study took place at the beginning of 2016 and only scales before 2016 could be taken into account for scale development. The litera-
ture review was extended to January 2018 to make sure that more recent developments are also captured.
B. Kump et al. 7
Rashidirad et al., 2017) were derived from Teece et al.’s (1997) theoretical assumptions, with only loose references
to Teece’s (2007) concept.
4. Scale development
4.1 Overall procedure
Even if the suggested approaches for scale development (Churchill, 1979; Hinkin, 1995, 1998) differ in their details,
they agree on the general process and quality criteria to be taken into account. We integrated their methodological
considerations and employed a three-step procedure, which will be described in detail below. First, in the item-gener-
ation step, we developed indicators that should reflect Teece’s (2007) sensing, seizing, and transforming capacities.
To provide a complementary perspective to Wilden et al.’s (2013) scale, our items should focus on the general exist-
ence of DC routines and outcomes (instead of frequencies of DC activities).
8 Toward a dynamic capabilities scale
We considered to start from existing (sub-)scales and combine them into one new instrument that measures all of
the aspects suggested by Teece (2007). However, we encountered two obstacles, one conceptual and one methodical.
4.2.2 Operationalization of DC
Sensing. As outlined in the theoretical considerations above, we posit that an organization with high sensing capacity
is able to systematically, continuously, and reliably acquire strategically relevant information from the environment,
including market trends, best practices, and competitors’ activities, that is, information from outside the organiza-
tion. Related concepts to sensing capacity have used the terms knowledge acquisition (Jantunen, 2005) or
B. Kump et al. 9
environmental scanning (Danneels, 2008). The systematic monitoring of the environment increases the chances of
becoming aware of upcoming markets, trends, and technology developments and of tapping into new business areas
Seizing. Further concretizing Teece’s (2007) definition, we assume that the capacity for seizing within an organiza-
tion is high if it is able to decide whether some information is of potential value, to transform valuable information
into concrete business opportunities that fit the organization’s strengths and weaknesses, and to make decisions ac-
cordingly. To operationalize seizing, we built on the content of existing scales but, for several reasons, could not dir-
ectly reuse existing items: Naldi et al. (2014) do not use actual survey items but other proxies (e.g., number of
newsletters a firm subscribed to). Jantunen (2005) and Makkonen et al. (2014) measure seizing but combine it with
sensing. Nedzinskas et al. (2013) employ items for seizing but do not report them. We also integrated contents from
scales for measuring absorptive capacity, such as Flatten et al. (2011), but adapted the items to the aspect of seizing.
To measure seizing, we used items such as “We recognize what new information can be utilized in our company,” or
“Our company is capable of turning new technological knowledge into process and product innovation.”
Transforming. In line with Teece’s (2007) ideas, we assume that an organization with a high transforming capacity
consistently implements decided renewal activities by assigning responsibilities, allocating resources, and ensuring
that the workforce possesses the newly required knowledge. To operationalize transforming capacity, we reviewed
existing reconfiguration and transforming (sub-)scales. Of the scales building on Teece’s model, Nedzinskas et al.
(2013) do not report items. Hawass’s (2010) reconfiguration scale focuses on the consequences of the transformation
rather than on the transformation itself (e.g., “We are more successful than competitors in diversifying into new mar-
kets by deploy [sic!] in existing technologies”; 426). Wilden and Gudergan (2015) ask for the frequency of how often
specific activities have been carried out (e.g., frequency of modifications in strategy and frequency of renewal of
business processes) as an indicator of transforming capacity. None of these scales explicitly asks for the capacity
for transforming processes and structures within an organization. One scale that measures transforming capacity
more directly is Li and Liu’s (2014) change implementation subscale. However, their items focus on conditions
for transforming (e.g., “We help each other in strategic change implementation”), not on its outcome. Our opera-
tionalizations of transforming capacity focus on how successful strategic renewal is actually implemented
and achieved within the organization and includes items such as “Decisions on planned changes are pursued con-
sistently in our company,” or “By defining clear responsibilities, we successfully implement plans for changes in
our company.”
Table 2. Pattern matrix of items for measuring sensing (SE), seizing (SZ), and transforming (T) capacities (EFA)
SE1 Our company knows the best practices in the market 0.72
SE2 Our company is up-to-date on the current market situation 0.82
SE3 Our company systematically searches for information on the current market situation 0.95
SE4 As a company, we know how to access new information 0.83
SE5 Our company always has an eye on our competitors’ activities 0.70
SE6a Our company quickly notices changes in the market 0.40 0.48
SZ1 Our company can quickly relate to new knowledge from the outside 0.87
SZ2 We recognize what new information can be utilized in our company 0.71
SZ3 Our company is capable of turning new technological knowledge into process and product innovation 0.84
SZ4 Current information leads to the development of new products or services 0.73
T1 By defining clear responsibilities, we successfully implement plans for changes in our company 0.89
T2 Even when unforeseen interruptions occur, change projects are seen through consistently in our company 0.90
T3 Decisions on planned changes are pursued consistently in our company 0.61
T4 In the past, we have demonstrated our strengths in implementing changes 0.60
T5 In our company, change projects can be put into practice alongside the daily business 0.72
T6a In our company, plans for change can be flexibly adapted to the current situation 0.44 0.55
Note: Extraction method: principal component analysis; rotation method: oblique rotation (promax) with Kaiser normalization; the rotation converged in six iter-
ations. Factor loadings <0.30 are suppressed in the table.
a
Item removed from survey after explorative factor analysis due to factor cross-loadings.
while at the same time reflecting the DC dimension well. Regarding inter-rater agreement, only such items
were considered where all five researchers agreed that they reflected an important aspect of Teece’s original
concept.
The aims of Phase 2 were to augment content validity (Rossiter, 2008), enhance comprehensibility, and minimize
perceived redundancy of the items developed in the first phase. Therefore, we conducted a first systematic walk-
through with an experienced top-level manager from a large firm in the IT industry. In a simplified version of the cog-
nitive lab technique (Wilson, 2004), the manager was asked to express his spontaneous associations and to comment
on the process of choosing an answer to ensure comprehension, logic, and relevance. The results were documented
and built the basis of another round of discussion among the five researchers. In the course of this discussion, the
items were slightly rephrased based on the managers’ feedback regarding comprehensibility, thereby ensuring that
they were still in line with theoretical assumptions (e.g., “In our company, anything new is easily transformed into
ideas for changes” was rephrased into “Our company is capable of turning new technological knowledge into process
and product innovation”). Moreover, items that were asking for the same information according to the manager
were reconsidered and merged, if necessary, reducing the overall number of items to 19. Then, a second walkthrough
was carried out with another experienced manager, the CEO of a small business, who again provided feedback on
the items by “thinking aloud.” Also, the second manager’s responses were documented for further discussion among
the team of researchers. The subsequent final round of discussion led to further slight rephrasing, and a reduction of
items that were asking for the same information (e.g., “We do not miss new developments that could have an influ-
ence on our business” and “Our company is up-to-date on the current market situation”), resulting in a “final” pre-
liminary scale of 16 items.
purify the scale, if needed. At this stage, the following test-theoretical considerations were underlying our sampling
strategy: (i) the scale should basically be applicable to firms in all industries; (ii) the scale should be applicable regard-
Variables. We developed an electronic questionnaire which included the 16 items of the three subscales sensing, seiz-
ing, and transforming (see Table 2). Moreover, we queried the firm size, firm age, position of the respondents, and in-
dustry at the end of the questionnaire.
Setting and respondents. For the pilot study, we had the opportunity to distribute an electronic version of the prelim-
inary scale to 100 CEOs of randomly selected firms (50 small and 50 large firms) in each of the seven sections of the
Austrian Federal Economic Chamber (i.e., in total 700 firms). Each recipient in this pilot study was asked to complete
the survey and to forward it to other persons in leading positions in their own industry network. This snowball-
sampling technique should increase the number of responses within sectors by invitation of other CEOs through their
peers. While the drawback was that we could not track the number of forwards to other respondents, the snowball-
sampling was efficient, as it revealed 269 complete responses.
Of these 269 questionnaires, 29.6% had been completed by CEOs, 35.9% by senior executives, and the
remaining 34.6% by other staff members in leading positions. Approximately half (51.1%) of the companies
had fewer than 250 employees. Most respondents (79.4%) worked for a company that was 10 or more years
old. The distribution within the industrial sectors of the Austrian Federal Economic Chamber of those who indi-
cated their industries was as follows: 11.0% bank and insurance, 13.3% crafts and trades, 11.9% commerce,
22.5% industry, 25.7% information and consulting, 11.9% tourism and leisure industries, and 3.7% transpor-
tation and communications. Overall, the goal of the sampling strategy to maximize variety across firm sizes and
industries was achieved.
Table 3. Descriptives, correlations and alpha coefficients for sensing (SE), seizing (SZ), and transforming (T) subscales
(n ¼ 269)
Setting and respondents. To reach firms from innovative industries, we reviewed the standard industry classification
in the European Union, Nomenclature statistique des Activités économiques dans la Communauté Européenne
(NACE) together with the results of the Eurostat community innovation survey (CIS) 2014 (Eurostat, 2014) and
selected subclasses of industries that are facing rather high environmental dynamism and innovative enterprises,
namely, Classes C–K, M, and Q. Based on contact information in the standard Austrian firm database Aurelia, we
B. Kump et al. 13
contacted 5229 companies within these subsections via e-mail. We specifically addressed the CEOs or middle-line
managers responsible for innovation issues.
coefficients for subscale intercorrelations are high (ranging from 0.50 to 0.70), indicating that the subscales for the
sensing, seizing, and transforming capacities measure distinct, but strongly related constructs.
Potential method biases. Even though we were striving for maximum rigor during the process of scale development,
our approach to data collection contains the possibility of biases due to systematic measurement error, because our
findings are based on self-reported measures, and all information for one case (i.e., one firm) stems from one re-
spondent (Podsakoff et al., 2003). To test for common method bias, we employed a CFA marker technique
(Richardson et al. 2009; Williams et al., 2010) based on structural equation modeling, with IO as marker variable.
One major advantage of this technique is that it allows evaluating method variance on both model- and item-level. In
the marker-based procedure, a baseline model, where the method factor loadings are forced to be zero (i.e., a model
without method factor loadings), is compared with a constrained model, where the method factor loadings are con-
strained to have equal values, and an unconstrained model, where the method factor loadings are estimated. The re-
sult of the v2 difference test between the baseline model and the constrained model reveals significant differences
(v2dif ¼ 98.67, df ¼ 1; P < 0.000, n ¼ 187), indicating the existence of a method factor. The comparison of the con-
strained model and the unconstrained model (v2dif ¼ 147.10, df ¼ 14; P < 0.000, n ¼ 187) shows that the impact of
the marker variable is not equal across items (loadings are ranging between 0.18 and 0.38). Of the three models, the
unconstrained model accounts best for marker variance related to the items of the scale (model fit: v2 ¼ 191.44,
df ¼ 126; v2/df ¼ 1.52, P < 0.000, CFI ¼ 0.97, RMSEA ¼ 0.05, n ¼ 187). Despite the method factor in the uncon-
strained model, all items also load significantly on the DC constructs (sensing, seizing, and transforming) they intend
to measure, and the model yields a good overall model fit. The median amount of method variance attributable to
the marker variable is 9, but the item loadings on the DC dimensions are still very high (ranging from 0.60 to 0.80).
Altogether, these results seem more than acceptable, as measures in psychology, sociology, marketing, business, and
education have been found to contain on average even about one quarter (26%) of method variance (Cote and
Buckley, 1987).
To test for nonresponse bias, we carried out a median split of early and late respondents (independent variable)
and calculated t-tests with sensing, seizing, and transforming (dependent variables). The tests revealed nonsignificant
effects for sensing and seizing, and a significant effect of small effect size (Cohen’s d ¼ 0.21; Cohen, 1988) for trans-
forming (t ¼ 1.99; P < 0.05), indicating that early respondents had slightly higher values (M ¼ 4.45; SD ¼ 0.85) than
late respondents (M ¼ 4.28; SD ¼ 0.80) as regards this one subscale. From these findings, we conclude that there is
no overall systematic nonresponse bias.
Criterion validity. To test the criterion validity of our scale, we employed the mean values of the sensing, seizing, and
transforming capacities as predictors of business performance and innovation performance in linear regression mod-
els (for descriptive statistics, see Table 4). We conducted an ordinary least squares regression analysis (Gelman and
Hill, 2007) to control for demographic variables (i.e., firm age and firm size) and the partial correlations between the
dimensions sensing, seizing, and transforming. In all calculations, we entered the demographic variables in the first
block and the dimensions of DC in the second.
As Table 5 shows, the regression model for business performance explains 33% (adjusted R2) of the variance in
the aggregate index of business performance. It also explains substantial variance of the distinct aspects of business
performance, namely, 30% of market performance, 17% of financial performance, 29% of employee-related per-
formance, and 17% of customer-related performance.
Of the demographic variables, the firm’s age has a significant, slightly negative effect on aggregate (b ¼ 0.04;
P < 0.01) market (b ¼ 0.06; P < 0.01) and financial performance (b ¼ 0.06; P < 0.05), indicating that as the firm’s
B. Kump et al. 15
Note: Model fit indices are NFI ¼ 0.94, CFI ¼ 0.97, IFI ¼ 0.97, TLI ¼ 0.96, and SRMR ¼ 0.04.
Table 4. Descriptive statistics and correlations for DC, and business and innovation performance indicators (n ¼ 307)
DC 0.91
Sensing (SE) 4.54 0.837 0.84 1
Seizing (SZ) 4.33 0.877 0.84 0.59** 1
Transforming (T) 4.31 0.876 0.87 0.50** 0.70** 1
Business 0.90
performance
Market 4.16 0.814 0.86 0.43** 0.48** 0.48** 1
Financial 3.89 1.053 0.91 0.36** 0.37** 0.33** 0.60** 1
Employee-related 4.53 0.947 0.83 0.35** 0.48** 0.52** 0.62** 0.42** 1
Customer-related 4.66 0.895 0.82 0.30** 0.37** 0.36** 0.82** 0.42** 0.64** 1
Innovation 0.80
performance
I1 2.60 1.994 0.07 0.27** 0.21** 0.17** 0.10 0.10 0.05 1
I2 2.17 1.935 0.06 0.25** 0.19** 0.18** 0.15** 0.13* 0.04 0.80** 1
I3 3.36 1.167 0.18** 0.36** 0.25** 0.26** 0.19** 0.15** 0.08 0.55** 0.45** 1
I4 2.25 1.627 0.16** 0.38** 0.30** 0.29** 0.19** 0.19** 0.15** 0.50** 0.45** 0.46** 1
I5 2.66 1.285 0.30** 0.35** 0.30** 0.37** 0.33** 0.19** 0.23** 0.24** 0.26** 0.33** 0.37** 1
age increases, its overall, market, and financial performance decrease in comparison with its most important compet-
itors. The firm’s size has a statistically significant, slightly negative effect on customer-related performance
5. Discussion
In the previous sections, we have documented our endeavor to systematically develop a scale for measuring DC as
conceptualized by Teece (2007). We will now discuss (i) the psychometric quality of the developed scale, (ii) implica-
tions for developing the concept of DC further, and (iii) benefits and drawbacks of our scale compared to other
instruments.
Table 5. Regression analysis for business performance and innovation performance variables (n ¼ 307)
(Constant) 1.88*** 1.58*** 1.73*** 2.75*** 1.98*** 2.00*** 1.60* 1.45** 0.60 0.14
Firm size 0.01 (0.03) 0.04 (0.04) 0.04 (0.03) 0.06* (0.03) 0.01 (0.02) 0.15* (0.07) 0.11(0.07) 0.07 (0.04) 0.09 (0.06) 0.14** (0.04)
Firm age 0.06** (0.02) 0.06* (0.03) 0.01 (0.02) 0.02 (0.03) 0.04* (0.02) 0.30*** (0.06)0.24*** (0.07)0.02 (0.03) 0.21*** (0.05)0.12*** (0.04)
Sensing (SE) 0.20** (0.06) 0.26** (0.08) 0.10 (0.07) 0.15* (0.07) 0.18** (0.06) 0.34* (0.16) 0.33* (0.16) 0.11 (0.10) 0.21(0.13) 0.14 (0.10)
Seizing (SZ) 0.17* (0.07) 0.20* (0.10) 0.21** (0.08) 0.18* (0.08) 0.19** (0.06) 0.63*** (0.18) 0.61*** (0.18) 0.51*** (0.11) 0.67*** (0.15) 0.29* (0.12)
Transforming (T) 0.23*** (0.06) 0.13 (0.10) 0.37*** (0.07) 0.18* (0.08) 0.23*** (0.06) 0.15 (0.17) 0.11 (0.17) 0.02 (0.10) 0.15 (0.14) 0.14(0.11)
R 0.56 0.44 0.55 0.43 0.58 0.40 0.36 0.38 0.46 0.43
R-squared 0.31 0.19 0.31 0.18 0.34 0.16 0.13 0.14 0.21 0.19
Adjusted 0.30 0.17 0.29 0.17 0.33 0.15 0.12 0.13 0.20 0.17
R-squared
coefficients) in two different samples. Validity was also high; the theoretically assumed three-factor structure was
identified in the EFA (scale purification step) and confirmed with the cross-validation in a new sample (scale confirm-
expenditure in R&D. Among the DC facets, seizing had the highest predictive power with regard to all measured per-
formance indicators. This seems plausible, as seizing is the capacity that is most closely linked to a strong strategy
structure may require (and thus indicate) transforming capacity; nevertheless, as Arend and Bromiley (2009) argued,
not changing does not necessarily imply an incapacity to change.
References
Agarwal, R. and C. E. Helfat (2009), ‘Strategic renewal of organizations,’ Organization Science, 20(2), 281–293.
Agarwal, R. and W. Selen (2009), ‘Dynamic capability building in service value networks for achieving service innovation,’ Decision
Sciences, 40(3), 431–475.
Ali, S., L. D. Peters and F. Lettice (2012), ‘An organizational learning perspective on conceptualizing dynamic and substantive capa-
bilities,’ Journal of Strategic Marketing, 20(7), 589–607.
Allred, C. R., S. E. Fawcett, C. Wallin and G. M. Magnan (2011), ‘A Dynamic Collaboration Capability as a source of competitive ad-
vantage,’ Decision Sciences, 42(1), 129–161.
Ambrosini, V. and C. Bowman (2009), ‘What are dynamic capabilities and are they a useful construct in strategic management?,’
International Journal of Management Reviews, 11(1), 29–49.
Arend, R. J. and P. Bromiley (2009), ‘Assessing the dynamic capabilities view: spare change, everyone?,’ Strategic Organization, 7(1),
75–90.
Babelyt_e-Labanausk_e, K. and S. Nedzinskas (2017), ‘Dynamic capabilities and their impact on research organizations’ R&D and in-
novation performance,’ Journal of Modelling in Management, 12(4), 603–630.
B. Kump et al. 21
Bagozzi, R. P. and H. Baumgartner (1994), ‘The evaluation of structural equation models and hypothesis testing,’ in R. P. Bagozzi
and H. Baumgartner (eds), Principles of Marketing Research. Blackwell Publishers: Cambridge, pp. 386–422.
Barrales-Molina, V., F. J. L. Montes and L. J. Gutierrez-Gutierrez (2015), ‘Dynamic capabilities, human resources and operating rou-
Gelman, A. and J. Hill (2007), Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge University Press:
New York.
Girod, S. J. G. and R. Whittington (2017), ‘Reconfiguration, Restructuring and Firm Performance: dynamic Capabilities and
Mitrega, M.,. D. Forkmann, G. Zaefarian and S. C. Henneberg (2017), ‘Networking capability in supplier relationships and its im-
pact on product innovation and firm performance,’ International Journal of Operations & Production Management, 37(5),
577–606.
Verreynne, M.-L., D. Hine, L. Coote and R. Parke (2016), ‘Building a scale for dynamic learning capabilities: the role of resources,
learning, competitive intent and routine patterning,’ Journal of Business Research, 69(10), 4287–4303.
Vickery, S. K., X. Koufteros and C. Droge (2013), ‘Does product platform strategy mediate the effects of supply chain integration on