ESWA 9540 No.

of Pages 11, Model 5G

18 September 2014

Expert Systems with Applications xxx (2014) xxxxxx

1

Contents lists available at ScienceDirect

Expert Systems with Applications

journal homepage: www.elsevier.com/locate/eswa

5
6

3 Knowledge management capabilities and rm performance: A test

4 of universalistic, contingency and complementarity perspectives
7 Q1 Jason F. Cohen , Karen Olsen
8 University of the Witwatersrand, Johannesburg, Private Bag X3, WITS, 2050, South Africa

9
10
a r t i c l e i n f o a b s t r a c t
1
2 2
3
13 Article history: Competing theoretical perspectives regarding the effects of knowledge management (KM) on perfor- 24
14 Available online xxxx mance have underpinned past empirical studies. By explicitly surfacing and comparing three such per- 25
spectives, we contribute to the theoretical advancement of the KM eld. We develop hypotheses 26
15 Keywords: consistent with the underlying logics of universalistic, complementarity and contingency theories and 27
16 Knowledge management we empirically test these hypotheses to determine which is best supported. Data was collected from a 28
17 Human capital sample of hospitality services rms operating in South Africa. Our results show that the universalistic 29
18 Firm performance
perspective is less preferred. We nd support for the complementarity perspective by revealing that cod- 30
19 Contingency
20 Complementarity
ication and human capital KM capabilities interact to inuence customer service outcomes. The contin- 31
21 Universalistic gency perspective also received support as the links between KM capabilities and performance were 32
22 found to be contingent on the business strategy of the rm. Our results suggest that future researchers 33
should explicitly acknowledge the theoretical perspective from which they are observing the perfor- 34
mance impacts of KM and ensure that empirical tests are consistent with the logic of the selected 35
perspective. 36
2014 Elsevier Ltd. All rights reserved. 37
38

39
40
41 1. Introduction human capital KM more important in others (e.g. Chen, Yeh, & 61
Huang, 2012; Greiner, Bhmann, & Krcmar, 2007). 62
42 Researchers have devoted much attention to empirical exami- Unfortunately, the growing volume of empirical research into 63
43 nation of the link between knowledge management (KM) and rm KM has not provided an answer as to which of the three perspec- 64
44 performance. Efforts have typically concentrated on the KM capa- tives offers a better explanation for the performance implications 65
45 bilities required for the externalization and codication of organi- of KM. As suggested elsewhere in the management literature 66
46 zational knowledge, and for the development and retention of tacit (Chnevert & Tremblay, 2011; Delery & Doty, 1996), it is important 67
47 knowledge embedded in human capital. Although not always to the theoretical advancement of a eld that alternate perspec- 68
48 explicitly acknowledged, competing theoretical perspectives tives are explicitly surfaced and compared. To that end, this paper 69
49 regarding the inter-relationship between these two KM capabili- aims to contrast universalistic, complementarity and contingency 70
50 ties and their implications for performance have however under- perspectives on the KM-to-performance relationship. We develop 71
51 pinned this past work. Some researchers view codication and hypotheses consistent with the underlying logic of each perspec- 72
52 human capital oriented KM capabilities as independent predictors tive and we empirically test these hypotheses to determine which 73
53 of performance and imply that their effects are universal across is best supported. Our approach will provide important guidance 74
54 operating contexts (e.g. Andreeva & Kianto, 2012; Wang & Wang, for future research efforts. 75
55 2012). Others view them not as independent but as complementary We select the South African hospitality services sector as the 76
56 and examine whether they act synergistically to effect perfor- empirical context for our study. Much past KM research has been 77
57 mance outcomes (e.g. Choi & Lee, 2003; Gloet & Terziovski, 2004; carried out in manufacturing and high-technology industries (e.g. 78
58 Storey & Hull, 2010). Another group of researchers takes a contin- Liu, Chen, & Tsai, 2004; Liu & Tsai, 2007; Marqus & Simn, 79
59 gency view (e.g. Hansen, Nohria, & Tierney, 1999), and nd codi- 2006) or within mixed-industries (Andreeva & Kianto, 2011; 80
60 cation more important to performance in certain contexts and Gold, Malhotra, & Segars, 2001; Lee, Lee, & Kang, 2005; 81
Sabherwal & Sabherwal, 2005). The effects of KM capabilities on 82
Corresponding author at: Department of Information Systems, Private Bag X3, the performance of rms in service sectors such as hospitality 83
WITS, 2050, South Africa. Tel.: +27 11 717 8164. has received less attention. Hospitality services are however recog- 84
E-mail address: jason.cohen@wits.ac.za (J.F. Cohen).

http://dx.doi.org/10.1016/j.eswa.2014.09.002
0957-4174/ 2014 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Cohen, J. F., & Olsen, K. Knowledge management capabilities and rm performance: A test of universalistic, contingency
and complementarity perspectives. Expert Systems with Applications (2014), http://dx.doi.org/10.1016/j.eswa.2014.09.002
 ESWA 9540 No. of Pages 11, Model 5G
18 September 2014

2 J.F. Cohen, K. Olsen / Expert Systems with Applications xxx (2014) xxxxxx

85 nized for their knowledge intensity. Because service delivery in have found similar effects. Lpez-Nicols and Meroo-Cerdn 149
86 hospitality occurs as a result of interaction between customers (2011) studied 310 Spanish rms from a mixed-industry sample 150
87 and employees, effective service delivery demands that employees and found codication and personalization to have independent 151
88 are knowledgeable about customer preferences and corresponding effects on multiple dimensions of organizational performance. Fur- 152
89 service procedures (Bouncken, 2002; Hallin & Marnburg, 2008). thermore, Wang and Wang (2012), in a study of 89 high-technol- 153
90 High rates of employee turnover, high percentages of unskilled ogy rms in China, found both explicit knowledge sharing 154
91 employees and changing customer preferences increases the pres- (codication) and tacit knowledge sharing independently associ- 155
92 sure on hospitality rms to improve their processes for transferring ated with various performance outcomes. 156
93 and saving knowledge, learning from employees experiences, and In contrast to the universalistic perspective, the complementar- 157
94 utilizing knowledge in service encounters (Bouncken, 2002). We ity perspective argues that the two KM capabilities are not inde- 158
95 therefore have an additional opportunity to extend theories of pendent but mutually reinforcing. They act synergistically and 159
96 KM into this high-potential but under-researched knowledge should be integrated into a more complete KM capability. This 160
97 context. view is grounded in recent extensions to the resource based view 161
98 The next section of this paper presents the conceptual back- of the rm, which contends that rm resources and capabilities 162
99 ground to our study. We discuss the two KM capabilities and con- can be congured into a complementary system where their joint 163
100 trast the universalistic, complementarity and contingency value is greater than the sum of their individual values (Tanriverdi 164
101 perspectives on how these two capabilities inuence performance. & Venkatraman, 2005). Bhatt (2001) acknowledged this perspec- 165
102 Drawing on the three perspectives, the studys hypotheses are then tive by suggesting that KM is best achieved through the interaction 166
103 developed. This is followed by a description of the research meth- of people and technological subsystems. Jasimuddin, Klein, and 167
104 ods, presentation of the empirical ndings and conclusions. Connell (2005) termed this a symbiosis strategy for KM. If a com- 168
plementarity exists then individual capabilities acting in isolation 169
105 2. Conceptual background will result in little performance gain while an integrated capability 170
would improve performance (Choi, Poon, & Davis, 2008; Tanriverdi 171
106 Codication and human capital approaches to KM are rooted in , 2006). Empirical ndings have supported this complementarity 172
107 the differences between explicit and tacit knowledge (Polanyi, perspective. For example, Choi and Lee (2003) found that explicit 173
108 1966). A knowledge codication capability is characterized by a and tacit methods interact to create a dynamic KM style that out- 174
109 coordinated managerial effort to externalize and formally repre- performs other styles that emphasize only one or the other. Addi- 175
110 sent the organizations explicit knowledge base (Hansen et al., tionally, Gloet and Terziovski (2004) concluded that innovation 176
111 1999; Rastogi, 2000). It is reected in formalized procedures for performance in manufacturing rms was dependent on an inte- 177
112 knowledge acquisition, and for converting and integrating grated approach of soft human capital and hard IT based KM 178
113 acquired knowledge, storing it in documents and computer sys- practices, whilst in the services context Storey and Hull (2010) 179
114 tems, making it usable and accessible, and embedding it into rou- found that rms with a combination KM strategy had the highest 180
115 tines and operating processes (Choi & Lee, 2003; Gold et al., 2001; levels of service innovation performance. 181
116 Greiner et al., 2007). Chong and Chong (2009) refer to this as a pro- Finally, the contingency perspective asserts that the relative 182
117 cess of knowledge construction, embodiment and deployment. importance of each KM capability depends on the operating con- 183
118 Information technology (IT) solutions are highly important to such text of the rm. This is rooted in contingency theorys proposition 184
119 codication efforts and play a predominant role in the acquisition, that performance is dependent upon the appropriate alignment or 185
120 storage and retrieval, protection, distribution and application of t of contextual factors with internal organization design 186
121 the rms knowledge (Alavi & Leidner, 2001; Bhatt 2001; (Zeithaml & Zeithaml, 1988). A rm will incur high costs attempt- 187
122 Davenport, De Long, & Beers, 1998; Gold et al., 2001). Codication ing to build and pursue a combined KM capability and will not 188
123 is associated with explicit-oriented (Choi & Lee, 2003) and IT par- enjoy the expected returns if codied and tacit knowledge have 189
124 adigms of KM (Gloet & Berrell, 2003). varying levels of importance to different types of rms. Firms 190
125 Human capital oriented KM emphasizes the value of tacit should therefore not look to reconcile the tensions between codi- 191
126 knowledge resident in the minds of the individual employees cation and human capital oriented KM but rather select between 192
127 who constitute the rms human capital. However, unlike IT-based the two approaches in a manner that best ts individual rm cir- 193
128 knowledge repositories, human capital is not owned by the rm cumstances (Hansen et al., 1999). The business strategy of the rm 194
129 (Bontis, Keow, & Richardson, 2000; Engstrm, Westnes, & represents one of the most important contextual factors underpin- 195
130 Westnes, 2003). Knowledge remains tied to individuals (Haesli & ning the contingency perspective where the alignment of KM to 196
131 Boxall, 2005), and is only available to the organization through business strategy is considered important for performance. For 197
132 employee willingness and motivation to contribute it (Zhou & example, Shih and Chiang (2005) found that cost leaders empha- 198
133 Fink, 2003). Therefore, human capital oriented KM focuses on size re-utilization of knowledge and lower the costs of knowledge 199
134 ensuring employee commitment and retention and successfully provision by implementing codication and IT-based KM, while 200
135 motivating employees to share their knowledge (Choi & Lee, companies pursuing differentiation strategies emphasize interac- 201
136 2002; Choi & Lee, 2003; Meso & Smith, 2000; Smedlund, 2008; tions and interpersonal connections among organization members 202
137 ajeva, 2010). This human capital oriented capability has varyingly for the creation of new knowledge. Furthermore, Truch and Bridger 203
138 been referred to as the personalization (Hansen et al., 1999), tacit- (2002) found that knowledge orientations vary across the business 204
139 oriented (Choi & Lee, 2003) and humanist paradigms of KM (Gloet strategy types of prospector, analyzer, defender and reactor, and 205
140 & Berrell, 2003). not all knowledge orientations are associated with performance 206
141 The universalistic perspective considers these two KM capabil- across all strategy types. Based on a multiple case study of 11 Ger- 207
142 ities as having independent effects on performance outcomes. man and Swiss companies, Greiner et al. (2007) concluded that 208
143 Devoting more effort to the management of codied knowledge that a business strategy focused on efciency requires a codica- 209
144 stocks and to the management of tacit knowledge and human tion strategy and a business strategy focused on innovation 210
145 capital should always be better than less effort for all rms. Recent requires a personalization KM strategy. Thus from the contingency 211
146 empirical studies support this perspective. For example, Andreeva perspective, organizations should emphasize either codication or 212
147 and Kianto (2012) found that IT and human capital based KM were human capital KM in a manner that aligns with their business 213
148 independently associated with the rms competitiveness. Others strategy. 214

Please cite this article in press as: Cohen, J. F., & Olsen, K. Knowledge management capabilities and rm performance: A test of universalistic, contingency
and complementarity perspectives. Expert Systems with Applications (2014), http://dx.doi.org/10.1016/j.eswa.2014.09.002
 ESWA 9540 No. of Pages 11, Model 5G
18 September 2014

J.F. Cohen, K. Olsen / Expert Systems with Applications xxx (2014) xxxxxx 3

215 3. Research hypotheses for service interactions to be driven in an orderly manner without 278
employees needing to constantly refer to other staff members 279
216 To achieve our objective of testing and comparing the three per- (Bouncken, 2002). Electronic databases, document repositories, 280
217 spectives, three alternative research models are derived and pre- and intranets ensure disparate information fragments can be con- 281
218 sented in Fig. 1. Given our empirical context in hospitality nected and large volumes of information stored, searched, and 282
219 services, we reect the dependent performance variables as cus- retrieved (Alavi & Leidner, 2001; De Carvalho & Ferreira, 2001; 283
220 tomer service outcomes and market and nancial performance. Lindvall, Rus, & Sinha, 2002; Sambamurthy & Subramani, 2005). 284
221 They are related in a manner predicted by the service-prot chain Decision support systems also embed knowledge-based decision 285
222 framework (Heskett, Jones, Loveman, Sasser, & Schlesinger, 1994). heuristics that facilitate quicker decision making and help trigger 286
223 The choice of these two outcomes is also consistent with prior action (Alavi & Leidner, 2001; De Carvalho & Ferreira, 2001). More- 287
224 work that has conrmed customer service, market, and nancial over, empirical evidence linking codication to rm performance is 288
225 performance as relevant to the evaluation of KM (Choy, Yew, & growing (e.g. Andreeva & Kianto, 2012; Gold et al., 2001; Liu et al., 289
226 Lin, 2006). The model depicts each KM capability as multi- 2004; Lpez-Nicols & Meroo-Cerdn, 2011; Sher & Lee, 2004; 290
227 dimensional. A codication capability is characterized by a set of Wang & Wang, 2012), and there have been recent calls to recognize 291
228 formalized codication practices and by the use of IT systems. A the value of codication and IT for KM within service sectors such 292
229 human capital KM capability is reected by the development and as hospitality (Hallin & Marnburg, 2008; Okumus, 2013). We can 293
230 retention of competent and committed employees, as well as therefore hypothesize that: 294
231 efforts to promote the sharing of tacit knowledge.
232 Model A illustrates the universalistic perspective by depicting Hypothesis 1. Codication KM capability is independently associ- 295
233 the two KM capabilities as having independent and unmoderated ated with increased rm performance outcomes. 296
234 effects on the performance of the rm. Model B reects the Human capital KM capability is reected by a rms success in 297
235 super-additive value synergies that arise from a complementarity developing and retaining competent and committed human capital 298
236 by depicting the two KM capabilities as interacting to affect perfor- and by its efforts to promote the sharing of tacit knowledge 299
237 mance. Modeling complementarities through interaction is consis- amongst employees. Through application of their tacit knowledge, 300
238 tent with prior studies (e.g. Song, Droge, Hanvanich, & Calantone, employees provide an organization with capabilities that compet- 301
239 2005). Finally, Model C depicts the contingency perspective by itors cannot easily copy (Teece, 2000). Firms with better human 302
240 modeling the business strategy of the rm as a contingency factor capital should be better positioned to implement performance 303
241 inuencing the strength of the relationships between the two KM enhancing process, product and service innovations. Such rms 304
242 capabilities and performance. The hypotheses underpinning the will be more exible and have better planning, problem solving 305
243 three models are expanded upon next. and troubleshooting capabilities, which can lead to improve cus- 306
tomer service and satisfaction (Youndt & Snell, 2004). The develop- 307
244 3.1. Universalistic perspective ment of human capital relies on the sharing of tacit knowledge 308
amongst employees, which provides learning opportunities and 309
245 Universalistic logic suggests that both a codication KM capa- helps to transfer skills, as well as motivate and empower employ- 310
246 bility and a human capital KM capability act independently and ees to improve their job performance (Hsu, 2008; Wang, Hult, 311
247 are important to performance regardless of the organizational Ketchen, & Ahmed, 2009). Additionally, knowledge sharing helps 312
248 context. rms to leverage the accumulated experiences of employees and 313
249 As indicated earlier, a codication KM capability is reected as a promotes teamwork in the resolution of problems and customer 314
250 set of formalized practices, supported by the use of IT systems for service needs (Hu, Horng, & Sun, 2009; Law & Ngai, 2008). More- 315
251 knowledge collection, documentation, storage and application. over, the interaction and communication that takes place during 316
252 Codication is considered valuable because poor employee com- knowledge sharing reduces uncertainty and can promote better 317
253 mitment and high rates of employee turnover pose signicant understanding and cooperation amongst employees, which results 318
254 threats to the rms knowledge base (Bontis et al., 2000). Without in a more positive working environment (Storey & Kelly, 2002). 319
255 a codication capability, the organizational knowledge base is vul- Empirical evidence from the hospitality context suggests that 320
256 nerable to individual employee loyalties and reliant on the success human capital is associated with increased customer satisfaction 321
257 of a complex and costly set of interventions needed to promote and performance (Chi & Gursoy, 2009; Kim, Kim, Park, Lee, & Jee, 322
258 tacit knowledge sharing, and to retain and secure the commitment 2012; Nieves & Haller, 2014), and that knowledge sharing amongst 323
259 of knowledgeable human capital (Haesli & Boxall, 2005; employees can positively inuence outcomes at organizational and 324
260 Jasimuddin et al., 2005). By cataloging and structuring the organi- individual levels (Hu et al., 2009; Kim & Lee, 2013; Yang, 2010). 325
261 zations knowledge, codication practices also decrease the com- Consequently, it can also be hypothesized that: 326
262 plexity and lower the cost of accessing and re-using the rms
263 knowledge base (Choi & Lee, 2002; Choi & Lee, 2003; Greiner Hypothesis 2. Human capital KM capability is independently 327
264 et al., 2007; Schulz & Jobe, 2001). The use of technologies further associated with increased rm performance outcomes. 328
265 overcomes the limitations of manual and human-intensive knowl-
266 edge systems to improve the efcacy of KM processes (Alavi &
267 Leidner, 2001; Gold et al., 2001; Tanriverdi, 2005; Wang, Klein, & 3.2. Complementarity perspective 329
268 Jiang, 2007). For example, IT can support service staff in the rapid
269 and wide-scale acquisition of knowledge during their interactions Fig. 1 (Model B) depicts the perspective that the two KM capa- 330
270 with customers (Bhnstedt, Scholl, Rensing, & Steinmetz, 2010; bilities are complementary. Complementary resources and capabil- 331
271 Lee & Choi, 2003), and can subsequently enhance customer rela- ities are mutually reinforcing and they act synergistically to yield 332
272 tionships by providing rms the ability to integrate that knowl- super-additive returns (Tanriverdi & Venkatraman, 2005; Wade & 333
273 edge and apply it at the point-of-service in a manner that Hulland, 2004). From a complementarity perspective, tacit and 334
274 improves the rms responsiveness to customer needs (Anand, explicit knowledge cannot be decoupled. The two KM capabilities 335
275 Ward, & Tatikonda, 2010). Automated workow systems use are thus mutually reinforcing and act synergistically to reduce 336
276 embedded knowledge to consistently implement best practice overall deciencies in organizational knowledge resources. For 337
277 operations into work routines (Alavi & Leidner, 2001), and allow example, the potential of tacit knowledge may not be realized if 338

Please cite this article in press as: Cohen, J. F., & Olsen, K. Knowledge management capabilities and rm performance: A test of universalistic, contingency
and complementarity perspectives. Expert Systems with Applications (2014), http://dx.doi.org/10.1016/j.eswa.2014.09.002
 ESWA 9540 No. of Pages 11, Model 5G
18 September 2014

4 J.F. Cohen, K. Olsen / Expert Systems with Applications xxx (2014) xxxxxx

Model A. Universalistic Perspective

IT Systems
Codification
KM
Codification Capability
Practices

Customer Market and

Service Financial
Outcomes Performance

HC Develop.
& Retention Human
Capital KM
Knowledge Capability
Sharing

Model B. Complementarity Perspective

IT Systems
Codification
KM
Codification Capability
Practices
Market and
Customer
HC Develop. Financial
Service
& Retention Performance
Human Outcomes
Capital KM
Knowledge
Capability
Sharing

Codification
X
Human
Capital KM

Model C. Contingency Perspective

IT Systems
Codification
KM
Codification Capability
Practices

Customer Market and

Service Financial
Outcomes Performance

HC Develop.
& Retention Human
Capital KM
Knowledge Capability
Sharing Business
Strategy

Fig. 1. Research models illustrating the universalistic, complementarity, and contingency perspectives.

339 knowledge cannot be rapidly transferred within the rm interaction for knowledge sharing do not exceed its cost (Choi & 343
340 (Jasimuddin et al., 2005). Technologies can thus complement a Lee, 2003; Storey & Hull, 2010). However, without human capital 344
341 human capital focus by ensuring that as the size and complexity development, technology solutions to KM may simply increase 345
342 of the organizations social network increases, the benets of social the information overload (Choi & Lee, 2003). Thus tacit knowledge 346

Please cite this article in press as: Cohen, J. F., & Olsen, K. Knowledge management capabilities and rm performance: A test of universalistic, contingency
and complementarity perspectives. Expert Systems with Applications (2014), http://dx.doi.org/10.1016/j.eswa.2014.09.002
 ESWA 9540 No. of Pages 11, Model 5G
18 September 2014

J.F. Cohen, K. Olsen / Expert Systems with Applications xxx (2014) xxxxxx 5

347 is needed in order for employees to make effective use of codica- performance amongst rms competing on cost, while human cap- 409
348 tion tools and create appropriate processes for their use (Storey & ital KM may be more strongly associated with performance 410
349 Hull, 2010). Bhatt (2001) similarly describes how the technical and amongst rms competing on service quality. With past empirical 411
350 human KM subsystems can interact in a manner that uses technol- studies also supporting this contingency (Greiner et al., 2007; 412
351 ogies to overcome physical constraints to knowledge management Hsu, 2008; Shih & Chiang, 2005), it can be hypothesized that: 413
352 and still provides rms the human capital capacity to give mean-
353 ing to organizational knowledge. A synergistic KM capability also Hypothesis 4a. The relationship between codication oriented 414
354 enables rms to exploit existing knowledge by using codication knowledge management capability and performance outcomes 415
355 to ease the process of knowledge storage, access and application, will be stronger amongst rms competing on cost than those 416
356 whilst simultaneously developing new knowledge, and discovering competing on quality. 417
357 better ways of operating through building unique, inimitable
358 human capital (Choi & Lee, 2003; Jasimuddin et al., 2005; Storey
Hypothesis 4b. The relationship between human capital oriented 418
359 & Hull, 2010). Under this complementarity perspective, it can be
knowledge management capability and performance outcomes 419
360 hypothesized that:
will be stronger amongst rms competing on quality than those 420
competing on cost. 421
361 Hypothesis 3. The complementarity of codication and human
362 capital KM capabilities is associated with increased rm perfor-
363 mance outcomes.
4. Research methods 422

4.1. Survey methodology and sampling 423

364 3.3. Contingency perspective

To test these alternate theoretical perspectives, we selected a 424

365 Fig. 1 (Model C) depicts the contingency perspective and
survey methodology. Unlike qualitative studies, the survey cannot 425
366 reects the business strategy of the rm as a contingency factor
provide for an in-depth analysis of organizational context sur- 426
367 that determines the relative importance of each of the KM capabil-
rounding knowledge management practice. However, the survey 427
368 ities. Within the services context, business strategy can be dened
methodology allows for the collection of data from a large sample 428
369 as the choice of the rm to create customer value by pursuing a
of organizations in the real-world eld setting. This can provide for 429
370 low-cost, standardized service offering or a high-quality, differen-
greater condence in the generalizability of results. Moreover, the 430
371 tiated service offering. While the former focuses on cost reduction
use of a structured questionnaire instrument facilitates the collec- 431
372 so as to produce services more efciently than competitors, the lat-
tion of the quantitative data needed for testing the studys research 432
373 ter focuses on the delivery of a more customized service experi-
hypotheses. Although the survey is reliant on self-reports of key 433
374 ence, the costs of which will be recovered through higher price
informants, it is appropriate when the phenomenon under consid- 434
375 premiums (OFarrell, Hitchens, & Moffat, 1993). Amongst low-cost
eration (i.e. KM capabilities) are not directly observable across the 435
376 service rms, high levels of tacit knowledge and unique expertise
large sample and when manipulation of variables is not possible. 436
377 may have little impact on their ability to implement competitive
The sampling frame consisted of 656 hospitality service provid- 437
378 strategies reliant on scale economies, standardization, and consis-
ers operating in South Africa. It was constructed from the AA 438
379 tency of service delivery (Sun, Aryee, & Law, 2007; Walsh, Enz, &
Accommodation Guide for Southern Africa and the Solomons 439
380 Canina, 2008). There are higher costs when employees are interact-
Guide, which are comprehensive publications of information on 440
381 ing and sharing expertise and skill (Choi & Lee, 2003). This may
hospitality and accommodation service providers such as hotels, 441
382 work against the objectives of low-cost rms for whom decreasing
motels, resorts and lodges in South Africa. The hotel owner, general 442
383 the costs associated with knowledge transfer and use would be an
manager, or equivalent, of each establishment was selected as the 443
384 important operating objective. Low-cost rms are also prone to
key informant. They were considered to be most knowledgeable 444
385 high rates of employee turnover and are considered less dependent
about the knowledge practices and performance of their 445
386 on human capital for strategy implementation. For these rms
establishments. 446
387 investments into human capital beyond that required to deliver a
388 low-cost offering would therefore be expensive and redundant.
389 Thus, knowledge systems within low-cost rms should promote 4.2. Measures 447
390 economies of scale by efciently transferring and diffusing best
391 practices and ensuring the exploitation and repeated use of proven A structured questionnaire was constructed to capture informa- 448
392 codied knowledge (Schulz & Jobe, 2001). The availability, effective tion from the key informant on each of the studys variables. The 449
393 dissemination, and reuse of codied knowledge can save time and questionnaire employed multi-item scales to measure each con- 450
394 money, as well as enhance the ability of the low-cost rms to offer struct. The performance outcome measures were adapted from 451
395 a standardized and consistent service experience. On the other Ottenbacher (2007). Customer service outcomes were measured 452
396 hand, rms competing on service quality will be more dependent as the accommodation providers comparative success in ensuring 453
397 on retaining human capital and more reliant on employees sharing customer satisfaction, retention, loyalty, and trust. Market and 454
398 tacit knowledge amongst each other. Efforts to externalize and nancial performance was measured as comparative performance 455
399 codify knowledge may add little value to service differentiation in the areas of room occupancy rates, growth, market share, sales, 456
400 efforts because tapping into a structured, codied, and reusable protability and revenue per available room. The scales were 457
401 knowledge base is unlikely to benet the generation of new and anchored from 1 = much worse than competitors to 7 = much bet- 458
402 innovative customer-specic solutions (Greiner et al., 2007). ter than competitors. 459
403 Instead, these rms will benet from employees applying their Codication KM capability was reected in two components, 460
404 tacit knowledge to build elements of difference into the design namely formalized codication practices and IT support for KM. 461
405 and delivery of the rms service offering (Walsh et al., 2008). Items were measured on a seven-point Likert scale. Five items 462
406 The high costs associated with a human-centric knowledge system were used to reect codication through practices such as docu- 463
407 would be offset by the price premium these rms charge their cus- mentation, integration, and cataloging of knowledge for easy 464
408 tomers. Thus codication may be more strongly associated with retrieval (Gold et al., 2001). A further ve items were adapted from 465

Please cite this article in press as: Cohen, J. F., & Olsen, K. Knowledge management capabilities and rm performance: A test of universalistic, contingency
and complementarity perspectives. Expert Systems with Applications (2014), http://dx.doi.org/10.1016/j.eswa.2014.09.002
 ESWA 9540 No. of Pages 11, Model 5G
18 September 2014

6 J.F. Cohen, K. Olsen / Expert Systems with Applications xxx (2014) xxxxxx

466 Chen, Tsou, and Huang (2009), Gold et al. (2001) and Lee et al. Table 1
467 (2005) to capture IT support for knowledge management. Items Sample prole (n = 112).

468 reected the use of IT to support the acquisition of knowledge; No. of respondents %
469 the embedding of knowledge in databases and decision support Respondent job title
470 systems; the use of Intranet and electronic bulletin boards; the Executive or general manager 53 47.3
471 use of IT systems that prompt action and recommend solutions Operations manager 44 39.3
472 to problems; and the use of IT to protect knowledge. Other manager e.g. HR, revenue 11 9.8
Owner manager 4 3.6
473 Human capital KM is focused on developing and retaining com-
474 petent and committed human capital, and on promoting the shar- Number of rooms
1550 22 19.6
475 ing of tacit knowledge. Development and retention of human 51100 36 32.1
476 capital was measured by asking respondents to provide a compar- 101200 32 28.6
477 ative assessment of their rms success in maintaining employee 201300 14 12.5
478 satisfaction, retaining competent employees, and keeping their 300+ 8 7.1
479 employee turnover lower than that of competitors (1 = much Age of property
480 worse than competitors to 7 = much better than competitors). Less than 10 years 28 25.0
1120 years 40 35.7
481 Knowledge sharing was measured using six items on a seven-point
2130 years 19 17.0
482 Likert scale. Items were adapted from Gold et al. (2001) and Wang 3150 years 13 11.6
483 et al. (2009) to reect the sharing of knowledge between employ- More than 50 years 12 10.7
484 ees both horizontally, across organizational units and vertically, up Star rating
485 and down the organizational hierarchy. 2-star 5 4.5
486 For testing the contingency perspective, we used the relevant 3-star 46 41.1
487 hospitality rms star rating as a measure of business strategy. This 4-star 52 46.4
5-star 9 8.0
488 is consistent with the approach of Sun et al. (2007: 567), and pro-
489 vides for an objective evaluation of the hospitality establishments
490 strategic orientation. Lower star establishments offer a basic
5.3. Measurement model 525
491 accommodation service at lower cost, and higher star establish-
492 ments offer higher quality accommodation services at a price pre-
Prior to testing the hypothesized relationships, a conrmatory 526
493 mium. As per Sun et al. (2007), lower star-rated rms (3 stars or
factor analysis was carried out to ensure adequate reliability, con- 527
494 less) were classied as competing on the basis of low-cost, and
vergent, and discriminant validity of the measurement model. The 528
495 higher star-rated establishments (4 and 5 stars) were classied
partial least squares (PLS) approach using SmartPLS 2.0.M3 was 529
496 as competing on the basis of service quality. Where data was not
used (Ringle, Wende, & Will, 2005). Results are reported in Table 2. 530
497 provided by respondents, secondary documentation was examined
Convergent validity was established through examination of the 531
498 to determine the establishments star rating.
average variance extracted (AVE) score for each factor. All factors 532
499 Finally, we asked questions about chain afliation, establish-
have AVEs in excess of 0.50 indicating that each construct explains 533
500 ment size, and age of property. Prior to administration, the ques-
more than 50% of the variance in its observed measures. Moreover, 534
501 tionnaire was pre-tested by nine hotel general managers.
all items exhibit high factor loadings on their respective constructs 535
(see Table 2), and no items load higher on constructs they are not 536
intended to measure. Discriminant validity was established by 537
502 5. Empirical results ensuring that the variance shared between two constructs was less 538
than the variance shared between a construct and its indicators. 539
503 5.1. Response prole This was achieved by examining the square root of each constructs 540
AVE to ensure it was greater than the correlation between that 541
504 A total of 148 responses were received, representing a response construct and other constructs in the model (see Table 3). Compos- 542
505 rate of 22.5%. However, after removing responses with missing ite reliability of the constructs is given by Fornell and Larckers 543
506 data on all required variables, 112 useable responses were avail- measure of internal consistency, all of which are well above the 544
507 able for analysis. The prole of respondents is shown in Table 1. recommended 0.70. Cronbachs alpha coefcients are similarly 545
508 Larger and smaller establishments were fairly well represented high. The analysis of the measurement model thus satised 546
509 with roughly half the sample consisting of establishments with requirements for reliability, convergent and discriminant validity 547
510 more than 100 rooms. The majority of respondents (75%) had been and tests of hypotheses could proceed. 548
511 in operation for at least 10 years, with both chain and non-chain
512 afliated entities well represented at 59% and 41%, respectively.
513 Based on their star-ratings, 51 establishments were classied as 5.4. Hypothesis testing 549
514 competing with a low-cost strategy whilst 61 were classied as
515 competing on service quality. Following conrmation of the measurement model, the PLS 550
approach was used to test the alternative structural models and 551
their underlying hypotheses. Although PLS based structural equa- 552
516 5.2. Common method bias tion modeling (SEM) lacks certain model t indices associated with 553
covariance-based SEM, a PLS models structural quality can be 554
517 Because data was collected from a single key informant, evaluated with respect to the R2 and Q2 statistics. These provide 555
518 Harmans one-factor test was used to rule out concerns over for assessments of the predictive power and predictive relevance 556
519 common method bias. The procedure involves carrying out an of the structural model. Higher R2 values provide condence in 557
520 unrotated principal components analysis of all scale items and the models capability to explain the endogenous latent variables, 558
521 examining whether the rst factor accounts for a majority of the and Q2 > 0 indicate that observed values are adequately recon- 559
522 variance. Results indicated that no single factor accounted for a structed by the model and its parameter estimates (Henseler, 560
523 majority of variance (only 35.7%). Thus common method bias does Ringle, & Sinkovics, 2009). In addition, the signicance of the struc- 561
524 not appear to be a concern in this study. tural paths, as specied in the model, is used to validate the theo- 562

Please cite this article in press as: Cohen, J. F., & Olsen, K. Knowledge management capabilities and rm performance: A test of universalistic, contingency
and complementarity perspectives. Expert Systems with Applications (2014), http://dx.doi.org/10.1016/j.eswa.2014.09.002
 ESWA 9540 No. of Pages 11, Model 5G
18 September 2014

J.F. Cohen, K. Olsen / Expert Systems with Applications xxx (2014) xxxxxx 7

Table 2
Q5 Conrmatory factor analysis and scale reliabilities.

Constructs and Items 1 2 3 4 5 6

IT support for knowledge management
In our organization, IT facilitates the acquisition of knowledge about our customers, suppliers and/or competitors 0.679 0.290 0.394 0.073 0.128 0.292
Our IT systems prompt us to take action and recommend solutions to problems 0.860 0.401 0.478 0.186 0.171 0.412
Knowledge is embedded in our databases and decision support systems 0.875 0.507 0.423 0.271 0.286 0.363
We developed information systems like Intranet and electronic bulletin boards to share information and knowledge 0.645 0.275 0.321 0.119 0.091 0.230
Our IT systems enable knowledge to be protected from unauthorized access 0.779 0.389 0.345 0.170 0.155 0.230
Knowledge codication practices
We encourage employees to document their experiences 0.389 0.708 0.542 0.335 0.240 0.356
In our organization, the knowledge of individuals is recorded in a structured way, so that others in the organization 0.395 0.858 0.559 0.252 0.210 0.359
may benet from it
In our organization, knowledge is presented in a standard way 0.274 0.777 0.510 0.258 0.185 0.249
In our organization, knowledge is cataloged for ease of retrieval 0.433 0.856 0.675 0.156 0.123 0.262
Our organization has processes for integrating knowledge from different sources 0.469 0.811 0.648 0.195 0.260 0.283
Knowledge sharing
Our organization has systems and venues for people to share their knowledge with others in the company 0.484 0.588 0.796 0.157 0.143 0.304
Our employees regularly share knowledge with their superiors 0.420 0.586 0.876 0.279 0.082 0.288
Our employees regularly share knowledge with their subordinates 0.368 0.636 0.852 0.154 0.081 0.250
Our employees regularly share ideas with other employees even if they are based in different departments 0.370 0.623 0.876 0.245 0.159 0.256
Our organization has processes for exchanging knowledge between individuals 0.384 0.635 0.832 0.275 0.109 0.337
Our organization promotes sharing of knowledge between work groups/teams 0.557 0.662 0.889 0.389 0.195 0.373
Human capital development and retention
Employee satisfaction 0.201 0.279 0.301 0.885 0.496 0.472
Employee turnover 0.085 0.169 0.183 0.865 0.483 0.272
Competencies of employees 0.265 0.328 0.291 0.870 0.655 0.475
Customer service outcomes
Customer satisfaction 0.211 0.236 0.133 0.620 0.910 0.511
Customer retention 0.113 0.163 0.077 0.529 0.867 0.472
Customer loyalty 0.210 0.233 0.160 0.578 0.931 0.571
Customer trust 0.270 0.299 0.179 0.555 0.886 0.569
Market and nancial performance
Room occupancy 0.337 0.210 0.169 0.438 0.569 0.821
New customers attracted 0.243 0.261 0.215 0.426 0.459 0.804
Market share growth 0.263 0.337 0.292 0.474 0.498 0.862
Protability (last 3 years) 0.434 0.405 0.434 0.398 0.490 0.863
Sales (last 3 years) 0.321 0.347 0.373 0.358 0.469 0.840
Revenue per available room 0.422 0.343 0.280 0.299 0.466 0.782
Average variance extracted 0.598 0.646 0.730 0.763 0.808 0.687
Composite reliability 0.880 0.901 0.942 0.906 0.944 0.930
Cronbachs a 0.830 0.863 0.926 0.847 0.921 0.909

Table 3
Correlation matrix and test of discriminant validity.

Mean (Std Dev) 1a 2 3 4 5 6

(1) IT support for knowledge management 5.38 (1.0) 0.773
(2) Knowledge codication practices 5.28 (1.1) 0.494 0.804
(3) Knowledge sharing 5.46 (1.1) 0.514 0.730 0.854
(4) Human capital development and retention 5.28 (1.1) 0.223 0.308 0.303 0.873
(5) Customer service outcomes 5.82 (0.9) 0.228 0.262 0.156 0.636 0.899
(6) Market and nancial performance 5.24 (0.9) 0.410 0.387 0.360 0.480 0.593 0.829
a
Square root of construct AVE appear on diagonals, off-diagonals are inter-construct correlations.

563 retically assumed relationships between constructs. Results are tomer service outcomes is signicant. The results show that the 578
564 reported in Table 4. In all analyses, we controlled for the effects total effects (direct and indirect) of both codication (b = 0.265, 579
565 of chain afliation and log of size (number of rooms) on perfor- p < 0.01) and human capital focused knowledge management 580
566 mance, and when testing the universalistic and contingency per- (b = 0.395, p < 0.001) on market and nancial performance are sig- 581
567 spective we also controlled for the effects of business strategy on nicant, thus providing support for hypotheses 1 and 2. The uni- 582
568 performance. versalistic model explains approximately 35% of the variation in 583
569 The two KM capabilities were modeled in the reective mode customer service outcomes and 45% in market and nancial 584
570 and we used the factor scores of the underlying rst-order dimen- performance. 585
571 sions as the respective manifest indicators (as illustrated in Fig. 1). To test the complementarity perspective (Fig. 1, Model B), we 586
572 To test the universalistic perspective depicted in Fig. 1, Model A used the approach recommended by Wilson (2010) to model the 587
573 (hypotheses 1 and 2), the two KM capabilities were independently synergistic effects of the KM complementarity as the interaction 588
574 associated with performance outcomes. Table 4 shows that the between codication and human capital KM. Results show that 589
575 path between codication focused knowledge management and the interaction term has signicant additional effects on customer 590
576 market and nancial performance is signicant, and the path service outcomes beyond the independent effects of the two KM 591
577 between human capital focused knowledge management and cus- capabilities (b = 0.254, p < 0.05). The difference between Model B 592

Please cite this article in press as: Cohen, J. F., & Olsen, K. Knowledge management capabilities and rm performance: A test of universalistic, contingency
and complementarity perspectives. Expert Systems with Applications (2014), http://dx.doi.org/10.1016/j.eswa.2014.09.002
 ESWA 9540 No. of Pages 11, Model 5G
18 September 2014

8 J.F. Cohen, K. Olsen / Expert Systems with Applications xxx (2014) xxxxxx

Table 4
PLS Results of Model Testing.

Model A Model B Model C low-cost Model C high-quality Model C difference

independent interaction providers (n = 51) providers (n = 61) between paths (t-values)
effects
Paths
Log size ? customer service outcomes 0.116 0.115 0.043 0.151 0.064
Log size ? market and nancial performance 0.017 0.017 0.056 0.114 0.088
Chain afliation ? customer service outcomes 0.036 0.013 0.041 0.019 0.460
Chain afliation ? market and nancial 0.024 0.025 0.148 0.062 1.595
performance
Business strategy ? customer service 0.002 0.006
outcomes
Business strategy ? market and nancial 0.061 0.060
performance
Customer service outcomes ? market and 0.444*** 0.440*** 0.325* 0.492*** 0.842
nancial performance
Codication KM ? customer service outcomes 0.023 0.071 0.347# 0.152 2.557*
Human capital KM ? customer service 0.585*** 0.560*** 0.356# 0.684*** 1.625
outcomes
Codication KM ? market and nancial 0.254** 0.257** 0.413** 0.161 1.386
performance
Human capital KM ? market and nancial 0.135 0.136 0.030 0.201 0.974
performance
Codication-human capital KM 0.254*
interaction ? customer service outcomes
Codication-human capital KM 0.012
interaction ? market and nancial
performance
Total effectsa
Codication KM ? market and nancial 0.265** 0.288** 0.526** 0.086 2.219*
performance
Human capital KM ? market and nancial 0.395*** 0.382*** 0.146 0.537*** 2.056*
performance
Predictive power of model
R2 customer service outcomes 0.355 0.417 0.366 0.450
Adj. R2 customer service outcomes 0.325 0.384 0.296 0.400
R2 market and nancial performance 0.458 0.458 0.502 0.462
Adj. R2 Market and nancial performance 0.427 0.422 0.434 0.402
Q2 cross-validated redundancyb
Customer service outcomes 0.257 0.295 0.239 0.352
Market and nancial performance 0.309 0.309 0.310 0.320
***
p < 0.001 (the signicance of the paths was determined by bootstrap resampling (5000 resamples), which is used to produce standard errors for calculating t-values with
degrees of freedom = n 1).
**
p < 0.01 (the signicance of the paths was determined by bootstrap resampling (5000 resamples), which is used to produce standard errors for calculating t-values with
degrees of freedom = n 1).
*
p < 0.05 (the signicance of the paths was determined by bootstrap resampling (5000 resamples), which is used to produce standard errors for calculating t-values with
degrees of freedom = n 1).
#
p < 0.010 (the signicance of the paths was determined by bootstrap resampling (5000 resamples), which is used to produce standard errors for calculating t-values with
degrees of freedom = n 1).
a
Total effects include direct and indirect effects.
b
Q2 values calculated via blindfolding procedure with omission distance = 6.

593 and Model A in relation to customer service outcomes as the the high-quality group (4 and 5 star-rated hotels). Results are 611
594 dependent variable can be examined through Cohens effect size shown in Table 4 for each of the two groups. The statistical signif- 612
595 formula: f2 = [0.417 0.355]/[1 0.417] = 0.106. This suggests a icance of the differences between the path coefcients in the two 613
596 small to moderate effect size. The interaction effect is graphically groups are given by t-values calculated using the formula outlined 614
597 depicted in Fig. 2 and illustrates that human capital KM capabilities in Chin (2004). Codication has a negative effect on customer ser- 615
598 are required to ensure that investments into codication based KM vice outcomes for rms competing on quality, and a positive effect 616
599 practices pay off. Amongst rms with a low human capital KM at the p < 0.10 level for rms competing on cost. This difference is 617
600 capability, a strong focus on codication based KM will have unin- signicant (t = 2.557, p < 0.01). Moreover, codication has a signif- 618
601 tended negative implications for performance. These results pro- icant effect (b = 0.413, p < 0.01) on market and nancial perfor- 619
602 vide support for the complementarity perspective (hypothesis 3) mance only amongst low-cost rms. This lends support to H4a. 620
603 where customer service outcomes is the criterion variable. For The effect of a human capital focus on customer service outcomes 621
604 market and nancial performance, there was no additional effect is stronger amongst high-quality rms (b = 0.684, p < 0.001) than 622
605 from the interaction term. low-cost rms (b = 0.356, p < 0.10), lending some support to H4b. 623
606 Finally, to test the contingency perspective depicted in Fig. 1, The direct and total effect of codication on the market and nan- 624
607 Model C (hypotheses 4a and 4b), we used subgroup analysis. This cial performance of rms is signicant only for the low-cost group 625
608 allows us to compare the paths between the two approaches to (b = 0.526, p < 0.01), whilst the total effect of a human capital focus 626
609 knowledge management and performance in the low-cost is signicant only for the high-quality group (b = 0.537, p < 0.001). 627
610 hospitality group (3 and lower star-rated hotels) with those in Taken together these results support a contingency perspective 628

Please cite this article in press as: Cohen, J. F., & Olsen, K. Knowledge management capabilities and rm performance: A test of universalistic, contingency
and complementarity perspectives. Expert Systems with Applications (2014), http://dx.doi.org/10.1016/j.eswa.2014.09.002
 ESWA 9540 No. of Pages 11, Model 5G
18 September 2014

J.F. Cohen, K. Olsen / Expert Systems with Applications xxx (2014) xxxxxx 9

629 that the business strategy of the rm moderates the strength of the This suggests that codication and human capital KM can interact 670
630 effects of the two knowledge management capabilities on to exert greater effects on customer service outcomes than would 671
631 performance. be achieved by implementing the capabilities independently. Our 672
results thus provide added support to prior studies (Choi & Lee, 673

632 6. Discussion 2003; Gloet & Terziovski, 2004) who found a combination KM 674
capability important to performance. Our results also contrast 675

633 The objective of this study was to determine which of the uni- those of Choi et al. (2008) who in their study of 131 Korean rms 676

634 versalistic, complementarity, or contingency perspective offers a did not nd a complementarity relationship between explicit and 677

635 better explanation for the performance implications of two KM tacit oriented KM. They concluded that implementing explicit 678

636 capabilities, namely codication and human capital KM. The uni- and tacit oriented strategies together would decrease organiza- 679

637 versalistic assumption is that the two KM capabilities have inde- tional performance. However, we found that the complementarity 680

638 pendent and unmoderated effects on performance outcomes perspective can be supported when customer service is the out- 681

639 across all rms. This is in contrast to the contingency view that come of interest. The two KM capabilities are mutually reinforcing 682

640 the relative importance of each KM capability depends on the and act synergistically to bring improvements to customer service 683

641 business strategy of the rm. Furthermore, the complementarity by overcoming physical constraints to knowledge storage, access, 684

642 perspective does not consider the two KM capabilities as indepen- and application at the point of service whilst still providing the 685

643 dent but as mutually reinforcing and suggests they will interact to human capital needed to give meaning to stored knowledge and 686

644 inuence performance. to develop new knowledge (Bhatt, 2001; Choi & Lee, 2003; 687

645 Results conrm that the two KM capabilities are important to Jasimuddin et al., 2005). KM efforts must focus on tacit knowledge 688

646 rm performance. However, when predicting both customer ser- and human capital development to avoid information overload 689

647 vice outcomes and market and nancial performance outcomes, (Choi & Lee, 2003) and ensure employees can make appropriate 690

648 the complementarity and contingency models performed better use of codication tools (Storey & Hull, 2010). 691

649 than the universalistic model, and together they provide more Taken together, our results show that the universalistic per- 692

650 complete insights into how the two KM capabilities relate to spective is less preferred. It fails to consider the performance 693

651 performance. enhancements that result from the synergistic combination of 694

652 Codication KM does not have independent effects on perfor- codication and human capital KM capabilities. Moreover, it does 695

653 mance outcomes across all rms. Codication as a KM capability not consider that different KM capabilities are not equally effective 696

654 is most relevant amongst low-cost providers who can look to for- for all types of rms and that alignment of KM to business strategy, 697

655 malized practices and IT systems that help decrease the complexity as suggested by the contingency perspective, is important to 698

656 and lower the cost of accessing and using the knowledge needed performance. 699

657 for consistency in low-cost service operations (Choi & Lee, 2003;
658 Greiner et al., 2007; Sun et al., 2007; Walsh et al., 2008). 7. Conclusion and implications 700

659 Human capital KM, on the other hand, was found to have posi-
660 tive implications for customer service with its effects strongest Past research has not adequately addressed the question as to 701

661 amongst rms competing on service quality. It slightly larger total which amongst the universalistic, complementarity, or contingency 702

662 effect size within the full sample and positive effects within both perspectives offers the best explanation for empirical observations 703

663 subsamples appears consistent with suggestions that tacit knowl- of the link between KM and performance. Our study addressed 704

664 edge may have stronger overall effects on rm performance due this gap by comparing the three perspectives using data collected 705

665 to its unobservable and inimitable nature (Johannessen, Olaisen, from a sample of hospitality service providers. We found support 706

666 Q2 & Olsen, 2001; Teece, 2000). for the complementarity perspective when customer service was 707

667 Importantly, we found that the KM complementarity, modeled the dependent variable. However, our results do not rule out the 708

668 as the interaction between codication and human capital KM, contingency perspective, which was also strongly supported as 709

669 has signicant additional effects on customer service outcomes. the links between KM and performance were found to be contin- 710
gent on the business strategy of the rm. Specically, we have 711
shown that codication based approaches are more strongly related 712
to performance for hospitality providers competing on cost, whilst 713
a human capital focus is more strongly related to performance for 714
providers competing on quality. We also found that the perfor- 715
mance outcome of interest is an important consideration when 716
testing the impacts of KM. Moreover, we have shown that it is 717
important for researchers to explicitly acknowledge the theoretical 718
perspective underpinning their examination of KM and that 719
researchers need to ensure correspondence between empirical tests 720
and the underpinning logic of the selected perspective. 721
From a practical perspective, our results conrm that different 722
KM capabilities may not be equally effective under varying organi- 723
zational circumstances. Managers should prioritize knowledge 724
management practices in accordance with the strategic orientation 725
of their rm and in accordance with the desired performance out- 726
comes. For providers competing on service quality, human capital 727
rather than codied knowledge stored in IT systems is important to 728
performance, and over-investing in IT systems at the expense of 729
human capital development will not produce expected returns. 730
However, the two capabilities can be usefully integrated as a com- 731
plementarity to address knowledge requirements for effective cus- 732
Fig. 2. Codication  human capital KM interaction on customer service outcomes. tomer service delivery. 733

Please cite this article in press as: Cohen, J. F., & Olsen, K. Knowledge management capabilities and rm performance: A test of universalistic, contingency
and complementarity perspectives. Expert Systems with Applications (2014), http://dx.doi.org/10.1016/j.eswa.2014.09.002
 ESWA 9540 No. of Pages 11, Model 5G
18 September 2014

10 J.F. Cohen, K. Olsen / Expert Systems with Applications xxx (2014) xxxxxx

734 Some limitations of our study should be noted. First, this study Delery, J. E., & Doty, D. H. (1996). Modes of theorizing in strategic human resource 809
management: tests of universalistic, contingency, and congurational 810
735 was cross-sectional and therefore we cannot conrm causal associ- 811
performance predictions. Academy of Management Journal, 39(4), 802835.
736 ations. Future research may wish to adopt a longitudinal research Engstrm, T. E. J., Westnes, P., & Westnes, S. F. (2003). Evaluating intellectual capital 812
737 design and provide stronger evidence for causality. Second, our in the hotel industry. Journal of Intellectual Capital, 4(3), 287303. 813
Gloet, M., & Berrell, M. (2003). The dual paradigm nature of knowledge 814
738 empirical work relied on self-reported data provided by a single 815
management: implications for achieving quality outcomes in human resource
739 key informant. Although we tested for common-method bias, it management. Journal of Knowledge Management, 7(1), 7889. 816
740 still remains that results would have been improved through the Gloet, M., & Terziovski, M. (2004). Exploring the relationship between knowledge 817
management practices and innovation performance. Journal of Manufacturing 818
741 use of more objective performance measures (if such were avail-
Technology Management, 15(5), 402409. 819
742 able) and/or the use of multiple key informants. Third, our sample Gold, A. H., Malhotra, A., & Segars, A. H. (2001). Knowledge management: an 820
743 is restricted to service rms in the hospitality sector in South Africa organizational capabilities perspective. Journal of Management Information 821
744 and may limit the generalizability of our conclusions for other Systems, 18(1), 185214. 822
Greiner, M. E., Bhmann, T., & Krcmar, H. (2007). A strategy for knowledge 823
745 rms. Finally, our study reected each of the knowledge manage- management. Journal of Knowledge Management, 11(6), 315. 824
746 ment approaches through selected rst-order constructs. Future Haesli, A., & Boxall, P. (2005). When knowledge management meets HR strategy: An 825
747 studies may wish to identify other practices associated with each exploration of personalization-retention and codication-recruitment 826
congurations. International Journal of Human Resource Management, 16(11), 827
748 of the approaches and examine the extent to which they are asso- 19551975. 828
749 ciated with rm performance. Hallin, C. A., & Marnburg, E. (2008). Knowledge management in the hospitality 829
industry: A review of empirical research. Tourism Management, 29(2), 366381. 830
Hansen, M. T., Nohria, N., & Tierney, T. (1999). Whats your strategy for managing 831
750 Acknowledgement knowledge? Harvard Business Review, 77(2), 106116. 832
Henseler, J., Ringle, C. M., & Sinkovics, R. R. (2009). The use of partial least squares 833
751 Q3 This research work was supported in part by the National path modeling in international marketing. Advances in International Marketing, 834
20, 277319. 835
752 Q4 Research Foundation (NRF) of South Africa. 836
Heskett, J. L., Jones, T. O., Loveman, G. W., Sasser, W. E., Jr., & Schlesinger, L. A.
(1994). Putting the service-prot chain to work. Harvard Business Review, 72(2), 837
753 References 164174. 838
Hsu, I. (2008). Knowledge sharing practices as facilitating factor for improving 839
organizational performance through human capital: a preliminary study. 840
754 Alavi, M., & Leidner, D. E. (2001). Knowledge management and knowledge
Experts Systems with Applications, 35(3), 13161326. 841
755 management systems: conceptual foundations and research issues. MIS
Hu, M. M., Horng, J., & Sun, Y. C. (2009). Hospitality teams: knowledge sharing and 842
756 Quarterly, 25(1), 107136.
service innovation performance. Tourism Management, 30(1), 4150. 843
757 Anand, G., Ward, P. T., & Tatikonda, M. V. (2010). Role of explicit and tacit
Jasimuddin, S. M., Klein, J. H., & Connell, C. (2005). The paradox of using tacit and 844
758 knowledge in six sigma projects: An empirical examination of differential
explicit knowledge: strategies to face dilemmas. Management Decision, 43(1), 845
759 project success. Journal of Operations Management, 28(4), 303315.
102112. 846
760 Andreeva, T., & Kianto, A. (2011). Knowledge processes, knowledge-intensity and
Johannessen, J., Olaisen, J., & Olsen, B. (2001). Mismanagement of tacit knowledge: 847
761 innovation: a moderated mediation analysis. Journal of Knowledge Management,
the importance of tacit knowledge, the danger of information technology, and 848
762 15(6), 10161034.
what to do about it. International Journal of Information Management, 21(1), 849
763 Andreeva, T., & Kianto, A. (2012). Does knowledge management really matter?
320. 850
764 Linking knowledge management practices, competitiveness and economic
Kim, T. T., Kim, W. G., Park, S. S., Lee, G., & Jee, B. (2012). Intellectual capital and 851
765 performance. Journal of Knowledge Management, 16(4), 617636.
business performance. What structural relationships do they have in upper- 852
766 Bhatt, G. D. (2001). Knowledge management in organizations: examining the
scale hotels? International Journal of Tourism Research, 14(4), 391408. 853
767 interaction between technologies, techniques and people. Journal of Knowledge
Kim, T. T., & Lee, G. (2013). Hospitality employee knowledge-sharing behaviors in 854
768 Management, 5(1), 6875.
the relationship between goal orientations and service innovation behavior. 855
769 Bhnstedt, D., Scholl, P., Rensing, C., & Steinmetz, R. (2010). Enhancing an
International Journal of Hospitality Management, 34, 324337. 856
770 environment for knowledge acquisition based on web resources by automatic
Law, C. C. H., & Ngai, E. W. T. (2008). An empirical study of the effects of knowledge 857
771 tag type identication. In Proceedings of international conference on computer-
sharing and learning behaviors on rm performance. Expert Systems with 858
772 aided learning (pp. 380389). Hasselt, Belgium.
Applications, 34(4), 23422349. 859
773 Bontis, N., Keow, W. C. C., & Richardson, S. (2000). Intellectual capital and business
Lee, H., & Choi, B. (2003). Knowledge management enablers, processes and 860
774 performance in Malaysian industries. Journal of Intellectual Capital, 1(1), 85100.
organizational performance. An integrative view and empirical examination. 861
775 Bouncken, R. B. (2002). Knowledge Management for Quality Improvements in
Journal of Management Information Systems, 20(1), 179228. 862
776 Hotels. Journal of Quality Assurance in Hospitality and Tourism, 3(3/4), 2559.
Lee, K. C., Lee, S., & Kang, I. W. (2005). KMPI: Measuring knowledge management 863
777 Chen, J., Tsou, H. T., & Huang, A. Y. (2009). Service delivery innovation: antecedents
performance. Information & Management, 42(3), 469482. 864
778 and impact on rm performance. Journal of Service Research, 12(1), 3655.
Lindvall, M., Rus, I., & Sinha, S. (2002). Technology support for knowledge 865
779 Chen, Y. Y., Yeh, S. P., & Huang, H. L. (2012). Does knowledge management t
management. In Proceedings of advances in learning software organizations (pp. 866
780 matter to business performance? Journal of Knowledge Management, 16(5),
94103). Chicago, Illinois, USA. 867
781 671687.
Liu, P., Chen, W., & Tsai, C. (2004). An empirical study on the correlation between 868
782 Chnevert, D., & Tremblay, M. (2011). Between universality and contingency: An
knowledge management capability and competitiveness in Taiwans industries. 869
783 international study of compensation performance. International Journal of
Technovation, 24(12), 971977. 870
784 Manpower, 32(8), 856878.
Liu, P. L., & Tsai, C. H. (2007). Effect of knowledge management systems on 871
785 Chi, C. G., & Gursoy, D. (2009). Employee satisfaction, customer satisfaction, and
operating performance. An empirical study of hi-tech companies using the 872
786 nancial performance. An empirical examination. International Journal of
balanced scorecard approach. International Journal of Management, 24(4), 873
787 Hospitality Management, 28(2), 245253.
734. 874
788 Chin, W. (2004). Frequently asked questions-partial least squares and PLS-graph.
Lpez-Nicols, C., & Meroo-Cerdn, . L. (2011). Strategic knowledge management, 875
789 Available from <http://disc-nt.cba.uh.edu/chin/plsfaq.htm> Accessed January
innovation and performance. International Journal of Information Management, 876
790 2014.
31(6), 502509. 877
791 Choi, B., & Lee, H. (2002). Knowledge management strategy and its link to
Marqus, D. P., & Simn, F. J. G. (2006). The effect of knowledge management 878
792 knowledge creation process. Expert Systems with Applications, 23(3), 173187.
practices on rm performance. Journal of Knowledge Management, 10(3), 879
793 Choi, B., & Lee, H. (2003). An empirical investigation of KM styles and their effect on
143156. 880
794 corporate performance. Information & Management, 40(5), 403417.
Meso, P., & Smith, R. (2000). A resource-based view of organizational knowledge 881
795 Choi, B., Poon, S. K., & Davis, J. G. (2008). Effects of knowledge management strategy
management systems. Journal of Knowledge Management, 4(13), 224234. 882
796 on organizational performance. A complementarity theory-based approach.
Nieves, J., & Haller, S. (2014). Building dynamic capabilities through knowledge 883
797 Omega, 36(2), 235251.
resources. Tourism Management, 40, 224232. 884
798 Chong, C. W., & Chong, S. C. (2009). Knowledge management process effectiveness:
OFarrell, P. N., Hitchens, D. M., & Moffat, L. A. R. (1993). The competitive advantage 885
799 measurement of preliminary knowledge management implementation.
of business service rms: a matched pair analysis of the relationship between 886
800 Knowledge Management Research & Practice, 7(2), 142151.
generic strategy and performance. Service Industries Journal, 13(1), 4064. 887
801 Choy, C. S., Yew, W. K., & Lin, B. (2006). Criteria for measuring KM performance
Okumus, F. (2013). Facilitating knowledge management through information 888
802 outcomes in organisations. Industrial Management & Data Systems, 106(7),
technology in hospitality organizations. Journal of Hospitality and Tourism 889
803 917936.
Technology, 4(1), 6480. 890
804 Davenport, T. H., De Long, D. W., & Beers, M. C. (1998). Successful knowledge
Ottenbacher, M. C. (2007). Innovation management in the hospitality industry: 891
805 management projects. Sloan Management Review, 39(2), 4357.
Different strategies for achieving success. Journal of Hospitality and Tourism 892
806 De Carvalho, R. B., & Ferreira, M. A. T. (2001). Using information technology to
Research, 31(4), 431454. 893
807 support knowledge conversion processes. Information Research, 7(1). Available
Polanyi, M. (1966). The tacit dimension. London: Routledge. 894
808 from <http://InformationR.net/ir/7-1/paper118.html> Accessed January 2014.

Please cite this article in press as: Cohen, J. F., & Olsen, K. Knowledge management capabilities and rm performance: A test of universalistic, contingency
and complementarity perspectives. Expert Systems with Applications (2014), http://dx.doi.org/10.1016/j.eswa.2014.09.002
 ESWA 9540 No. of Pages 11, Model 5G
18 September 2014

J.F. Cohen, K. Olsen / Expert Systems with Applications xxx (2014) xxxxxx 11

895 Rastogi, P. N. (2000). Knowledge management and intellectual capital the virtuous Tanriverdi, H., & Venkatraman, N. (2005). Knowledge relatedness and the 933
896 reality of competitiveness. Human Systems Management, 19(1), 3948. performance of multibusiness rms. Strategic Management Journal, 26(2), 934
897 Ringle, C. M., Wende, S., & Will, A. (2005). SmartPLS. Release: 2.0.M3. Available from 97119. 935
898 <http://www.smartpls.de> Accessed January 2014. Teece, D. J. (2000). Strategies for managing knowledge assets: The role of rm 936
899 Sabherwal, R., & Sabherwal, S. (2005). Knowledge management using information structure and industrial context. Long Range Planning, 33(1), 3554. 937
900 technology: Determinants of short-term impact on rm value. Decision Sciences, Truch, E., & Bridger, D. A. (2002). The importance of strategic t in knowledge 938
901 36(4), 531567. management. In: European conference on information systems (ECIS) (pp. 905 939
902 ajeva, S. (2010). The analysis of key elements of socio-technical knowledge 918). Gdansk, Poland, June 68 2002. 940
903 management system. Economics and Management, 15, 765774. Wade, M., & Hulland, J. (2004). Review: The resource-based view and information 941
904 Sambamurthy, V., & Subramani, M. (2005). Special issue on information systems research: Review, extension, and suggestions for future research. MIS 942
905 technologies and knowledge management. MIS Quarterly, 29(1), 17. Quarterly, 28(1), 107142. 943
906 Schulz, M., & Jobe, L. A. (2001). Codication and tacitness as knowledge Walsh, K., Enz, C. A., & Canina, L. (2008). The impact of strategic orientation on 944
907 management strategies: An empirical exploration. Journal of High Technology intellectual capital investments in customer service rms. Journal of Service 945
908 Management Research, 12(1), 139165. Research, 10(4), 300317. 946
909 Sher, P. J., & Lee, V. C. (2004). Information technology as a facilitator for enhancing Wang, C. L., Hult, G. T. M., Ketchen, D. J., & Ahmed, P. K. (2009). Knowledge 947
910 dynamic capabilities through knowledge management. Information & management orientation, market orientation and rm performance. An 948
911 Management, 41(8), 933945. integration and empirical examination. Journal of Strategic Marketing, 17(2), 949
912 Shih, H. A., & Chiang, Y. H. (2005). Strategy alignment between HRM, KM, and 99122. 950
913 corporate development. International Journal of Manpower, 26(6), Wang, E., Klein, G., & Jiang, J. J. (2007). IT support in manufacturing rms for a 951
914 582603. knowledge management dynamic capability link to performance. International 952
915 Smedlund, A. (2008). The knowledge system of a rm: social capital for explicit, Journal of Production Research, 45(11), 24192434. 953
916 tacit and potential knowledge. Journal of Knowledge Management, 12(1), Wang, Z., & Wang, N. (2012). Knowledge sharing, innovation and rm performance. 954
917 6377. Expert Systems with Applications, 39(10), 88998908. 955
918 Song, M., Droge, C., Hanvanich, S., & Calantone, R. (2005). Marketing and technology Wilson, B. (2010). Using PLS to investigate interaction effects between higher order 956
919 resource complementarity: An analysis of their interaction effect in two branding constructs. In Handbook of partial least squares (pp. 621652). Berlin 957
920 environmental contexts. Strategic Management Journal, 26(3), 259276. Heidelberg: Springer. 958
921 Storey, C., & Hull, F. M. (2010). Service development success: A contingent approach Yang, J. T. (2010). Antecedents and consequences of knowledge sharing in 959
922 by knowledge strategy. Journal of Service Management, 21(2), 140161. international tourist hotels. International Journal of Hospitality Management, 960
923 Storey, C., & Kelly, D. (2002). Innovation in services: The need for knowledge 29(1), 4252. 961
924 management. Australasian Marketing Journal, 10(1), 5970. Youndt, M. A., & Snell, S. A. (2004). Human resource congurations, intellectual 962
925 Sun, L., Aryee, S., & Law, K. A. (2007). High-performance human resource practices, capital, and organizational performance. Journal of Managerial Issues, 16(3), 963
926 citizenship behavior, and organizational performance. A relational perspective. 337360. 964
927 Academy of Management Journal, 50(3), 558577. Zeithaml, V. A., & Zeithaml, C. P. (1988). The contingency approach: Its foundations 965
928 Tanriverdi, H. (2005). Information technology relatedness, knowledge management and relevance to theory building and research in marketing. European Journal of 966
929 capability and performance of multibusiness rms. MIS Quarterly, 29(2), Marketing, 22(7), 3764. 967
930 311334. Zhou, A. Z., & Fink, D. (2003). The intellectual capital web: A systematic linking of 968
931 Tanriverdi, H. (2006). Performance effects of information technology synergies in intellectual capital and knowledge management. Journal of Intellectual Capital, 969
932 multibusiness rms. MIS Quarterly, 30(1), 5777. 4(1), 3448. 970
971

Please cite this article in press as: Cohen, J. F., & Olsen, K. Knowledge management capabilities and rm performance: A test of universalistic, contingency
and complementarity perspectives. Expert Systems with Applications (2014), http://dx.doi.org/10.1016/j.eswa.2014.09.002