An Evaluation of Sourcing Strategies’ Relationship with Software Development Project Performance

Faculty of Engineering, Blekinge Institute of Technology, 371 79 Karlskrona, Sweden

An Evaluation of Sourcing Strategies’

Relationship with Software Development

Project Performance

Shenner Kotlji

Johanna Strand

Master of Science in Industrial Engineering and Management June 2020

i

This thesis is submitted to the Faculty of Engineering at Blekinge Institute of Technology in partial fulfilment of the requirements for the degree of Master of Science in Industrial Engineering and Management. The thesis is equivalent to 20 weeks of full-time studies.

The authors declare that they are the sole authors of this thesis and that they have not used any sources other than those listed in the bibliography and identified as references. They further declare that they have not submitted this thesis at any other institution to obtain a degree.

Contact Information:

Author(s):

Shenner Kotlji

shko14@student.bth.se Johanna Strand

jocj15@student.bth.se University advisor:

Anders Wrenne

Department of Industrial Economics

Faculty of Engineering

Blekinge Institute of Technology SE-371 79 Karlskrona, Sweden

Internet : www.bth.se Phone : +46 455 38 50 00 Fax : +46 455 38 50 57

ii

P

First and foremost, we would like to acknowledge and thank our university supervisor, Anders Wrenne, for his support and guidance during this journey, your insights have been truly invaluable and appreciated. We would also like to extend our gratitude to our external supervisor Dr Ricardo Britto for his desire to share valuable and unique knowledge on the field, it made us push our ideas and barriers. Bodil Jansson, your passion and support has been very motivating for us, thank you! We would also like to thank the project managers at the case firm who helped us with the data collection.

Five intense years of new perspectives, knowledge, and hard work have now come to an end.

For enthusiasm and encouragement, we would also like to direct our gratitude towards professors and lecturers at Blekinge Institute of Technology for helping out with the thesis and preparing us for the future.

For constant support, love, and kindness we would to express our deepest appreciation to our family and close friends for being present in the hardest of moments in this journey.

From the bottom of my heart, I would like to direct the deepest gratefulness towards my co-worker Shenner! Your constant passion for the field, ambition to succeed, and sharp mentality have been crucial for this thesis. Co-operating with you has been an absolute pleasure and I wish you all the greatest with your future commitments.

Johanna Strand

Blekinge Institute of Technology M.Sc. in Industrial Management and Engineering 300 ECTS

Master Thesis 30 ECTS 10^th June 2020.

_______________________________

I would like to express my appreciation to Johanna for this journey. Your constant drive, energy, and passion for our thesis have been truly inspiring. The encouragement and fruitful collaboration have been vital for the success of this thesis, thank you and best of luck in the future!

Shenner Kotlji

Blekinge Institute of Technology M.Sc. in Industrial Management and Engineering 300 ECTS

Master Thesis 30 ECTS 10^th June 2020.

_______________________________

iii

A BSTRACT

Background: The rapid improvement of technological infrastructure over the past three decades have led to increased connectivity and communication possibilities. This has allowed firms to develop sophisticated firm-specific governance structures with the use of different sourcing strategies. The main objectives have been to cut costs and gain competitive advantages by outsourcing activities offshore to developing countries or third-party vendors. The decision to outsource has its roots in the classical transaction cost theory and resource-based view. Some outsourcing partnerships have resulted in failures while others in success stories. The stories are evident in the software industry as the industry is highly susceptible to sourcing strategies, hence outsourcing is often used in global software development (GSD) projects. Firms engage in GSD with the ultimate goal of producing products faster, at a low-cost, and of high quality.

In spite of this, GSD projects face huge challenges in terms of geographical-, temporal-, and cultural distances while the added complexity and issues associated with different sourcing strategies amplifies these challenges. The use of different sourcing strategies in GSD projects has not received as much attention as the debate about whether co-located or distributed teams perform better.

Objective: The objective of this thesis is to explore how outsourcing, insourcing, and the combination of both in a co-located and distributed development setting relates to the software development project performance measured in terms of quality and productivity. The aim is to enrich and add to the scarce literature of global software development project performance in relation to sourcing strategies and when these are combined in projects.

Method: In this thesis, we employed a case study at a software firm which engages in large- scale global software development projects. A total of 64 projects were selected for the study and data was collected primarily from archival documents where we made use of management documents and code databases. The data analysis was conducted using statistical tests in SPSS to investigate relationships and differences in quality and productivity for the four strategies.

Results: The results revealed a statistically significant difference in quality among the sourcing strategies. Insourcing projects are associated with the highest quality followed by outsourcing projects and co-located mixed sourcing projects. Distributed mixed sourcing projects are generating the lowest quality. Although not statistically significant, distributed mixed sourcing projects are the most productive strategy followed by outsourcing projects, insourcing projects and lastly co-located mixed sourcing projects

Conclusions: The recent trends in GSD which indicate that many companies turn back to insourcing after outsourcing is justified in our results as insourcing projects displays the highest quality. While outsourcing projects have been associated with poor quality and productivity, our findings suggest they are very competitive in terms of both. The mix of internal employees and third-party consultants in a co-located and distributed setting is associated with lower quality while poor productivity is only attributed to the co-located case.

Keywords: Sourcing Strategies, Insourcing, Outsourcing, Global Software Development, Project Performance

iv

S AMMANFATTNING

Bakgrund: Teknikens framfart över de tre senaste decenierna har lett till ökade anslutnings- och kommunikationsmöjligheter. Detta har möjliggjort för företag att utveckla sofistikerade företagsspecifika stryelsestrukturer med använding av olika sourcing strategier. Det huvudsakliga målet med att outsourca olika aktiviteter till utvecklingsländer eller tredjepartsleverantörer har varit att minska kostnader och att erhålla konkrurrensfördelar.

Belsutet om att outsourca aktiviteter i ett företag är resursbaserat och grundar sig i den klassiska transaktionskostnadsteorin och resursbaserade teorin (resource-based view). Somliga partnerskap har resulterat i misslyckanden, andra i framgångssagor. Många av dessa har åstadkommits inom mjukvaruindustrin då industrin är mottaglig för olika sourcing strategier, bland annat används outsourcing mycket i globala mjukvaruutvecklingsprojekt. IT-företagen bedriver globala mjukvaruprojekt med de slutgiltliga målet att producera snabbare till ett lägre pris och en högre produktkvalitet. Trots detta uppstår ofta utmaningar när de gäller geografiska, tidsmässiga eller kulturella avstånd där nyttjandet av sourcing strategier adderar ytterligare ett lager av komplexitet. Majoriteten av uppmärksamhet har riktats mot hur samlokaliserade eller distrubuerade projekt presterar medan nyttjandet av sourcingstrategier i globala mjukvaruutvecklingsprojekt har fått mindre uppmärksamhet.

Syfte: Syftet är att undersöka hur outsourcing, insourcing och en kombination av de två i en samlokaliserad eller distrubuerad miljö relaterar till mjukvaruutvecklingsprojkets prestanda i termer om produktivitet och kvalitet. Målet är att berika och addera kunskap till en knapp litteratur om prestanda för globala mjukvaruutvecklingsprojekt med avseende på sourcing strategier när dessa tillämpas i en kombination.

Metod. En fallstudie har bedrivits på ett mjukvaruutvecklingsföretag som arbetar med globala storskaliga mjukvaruutvecklingsprojekt. Totalt har 64 projekts blivit utvalda, analyserade och grupperade i fyra olika sourcing strategier. Sekundärdata samlades primärt in från interna företagsarkiv såsom företagets interna databaser och projektledarverktyg. Intervjuer har också genomförts med två projektledare för att få en mer djupgående förståelse av projekten och för att välja ut projekten baserat på våra krav. Dataanalysen genomfördes via statistiska tester i SPSS.

Resultat: Resultatet visade en signifikant skilland i kvalitet mellan sourcing strategierna.

Insourcingprojekt visade sig bidra med högst kvalitet följt av outsourcing och mixad samlokaliserad sourcing projekt. Signifikat lägst kvalitet hade mixad distrubuerad sourcing.

Trots icke-signifikanta resultat, visade det sig att produktiviteten var som högst i mixad distribuerad sourcing projekt följt av outsourcing, mixad samlokaliserade sourcing och insourcing projekt.

Slutsats: De senaste trenderna inom global mjukvaruutveckling indikerar på att många företag återvänder till insourcing efter outsourcing är motiverade i vårt resultat eftersom insourcing projekt resulterar i högre kvalitet. Trots att outsourcingprojekt har förknippats med låg kvalitet och produktivitet tyder våra resultat på att de är mycket konkurrenskraftiga. Slutligen är mixen av internanställda och tredje part konsulter i en samlokaliserad och distrubuerad miljö förknippad med en sämre kvalitet, och i de samlokaliserade fallet även sämre produktivitet.

Nyckelord: Sourcing Strategier, Insourcing, Outsourcing, Global Mjukvaruutveckling, Projekt Performance

v

T ABLE OF C ONTENTS

ABSTRACT ... III SAMMANFATTNING ... IV TABLE OF CONTENTS ... V TABLE OF TABLES ... VII

TABLE OF FIGURES ... VIII

ABBREVIATIONS ... IX TERMINOLOGY ... X

1 INTRODUCTION ... 1

1.1 PROBLEM DISCUSSION ... 3

1.2 PURPOSE AND RESEARCH QUESTION ... 5

1.3 DELIMITATIONS ... 5

1.4 OUTLINE ... 6

2 THEORETICAL FRAMEWORK ... 7

2.1 STRATEGIC SOURCING AND GOVERNANCE STRUCTURE ... 7

2.1.1 Transaction cost theory ... 7

2.1.2 Resource-based view ... 8

2.2 GLOBAL SOFTWARE DEVELOPMENT ... 8

2.2.1 Drivers and challenges ... 8

2.3 SOURCING STRATEGIES IN ICT ... 10

2.3.1 Terminology for sourcing strategies ... 10

2.3.2 Project arrangements ... 11

2.4 PROJECT PRODUCTIVITY, QUALITY AND PERFORMANCE ... 12

2.4.1 Project performance in software development ... 13

2.5 THEORETICAL SUMMARY ... 14

3 RELATED WORK ... 15

3.1 BRIDGING THE GAP IN THE LITERATURE ... 18

4 METHOD ... 19

4.1 METHOD SELECTION ... 19

4.1.1 Case study ... 19

4.2 RESEARCH DESIGN ... 20

4.2.1 Overview ... 20

4.3 THE CASEAND UNIT OF ANALYSIS ... 21

4.4 OPERATIONALIZATION OF VARIABLES ... 22

4.4.1 Dependent variables ... 22

4.4.2 Independent variables ... 23

vi

4.4.3 Control variables ... 24

4.4.4 Hypotheses ... 24

4.4.5 Model specification ... 24

4.5 DATA COLLECTION ... 25

4.5.1 Archival research ... 26

4.5.2 Unstructured interviews ... 27

4.6 DATA ANALYSIS ... 28

4.6.1 Data preparation ... 28

4.6.2 Statistical analysis ... 28

4.7 THREATS TO VALIDITY... 29

4.7.1 Reliability ... 30

4.7.2 Internal validity ... 30

4.7.3 Construct validity ... 30

4.7.4 External validity ... 30

4.8 RESEARCH ETHICS ... 31

5 RESULTS ... 32

5.1 PROJECT CHARACTERISTICS ... 32

5.2 DIFFERENCES IN PERFORMANCE MEASURES BETWEEN SOURCING STRATEGIES ... 33

5.2.1 Productivity ... 33

5.2.2 Quality ... 34

5.3 HOW SOURCING STRATEGIES PREDICT PRODUCTIVITY, QUALITY AND THE EFFECT FROM OTHER CONTROL VARIABLES ... 36

6 ANALYSIS ... 39

6.1 PRODUCTIVITY ... 39

6.2 QUALITY... 40

7 DISCUSSION ... 43

7.1 THE TRADE-OFF BETWEEN SOURCING STRATEGIES AND PROJECT PERFORMANCE ... 43

7.2 THE MATURITY OF STRATEGIC PARTNERSHIPS ... 44

7.3 IMPLICATIONS FOR RESEARCH AND PRACTICE... 45

8 CONCLUSION ... 46

8.1 F^UTURE W^ORK ... 47

REFERENCES ... 48

APPENDICES ... 56

APPENDIX A ... 56

APPENDIX B ... 57

vii

T

T

TABLE 1.TAXONOMY FOR SOURCING STRATEGIES (ŠMITE ET AL.,2014). ... 10

TABLE 2.DATA SOURCES USED AND VARIABLES EXTRACTED FROM ARCHIVAL DOCUMENTS. ... 26

TABLE 3.SAMPLE SELECTION CRITERIA ... 27

TABLE 4.DESCRIPTIVE STATISTICS OF THE DATA COLLECTED FROM 64 PROJECTS. ... 32

TABLE 5.DESCRIPTIVE STATISTICS OVER SOURCING STRATEGIES AND PRODUCTIVITY ... 33

TABLE 6.DESCRIPTIVE STATISTICS OVER SOURCING STRATEGIES AND DEFECT DENSITY. ... 35

TABLE 7.MODEL SUMMARIES FOR PRODUCTIVITY AND QUALITY REGRESSION. ... 37

TABLE 8.BETA VALUES ARE SHOWN FOR THE PREDICTORS OF PRODUCTIVITY. ... 37

T^ABLE 9.BETA VALUES ARE SHOWN FOR THE PREDICTORS OF DEFECT DENSITY. ... 38

viii

T

F

FIGURE 1.A SUMMARY OF DRIVERS FOR GSD REPORTED FROM FIRMS IN EUROPE AND NORTH

AMERICA (EBERT ET AL.,2016). ... 9

FIGURE 2.PROJECT ARRANGEMENTS IN GSD CATEGORIZED BY SOURCING STRATEGIES, ADAPTED FROM ŠMITE (2014). ... 11

FIGURE 3.PURPOSED-BASED EFFECTIVENESS AND EFFICIENCY BY (WAGNER &DEISSENBOECK, 2019)(CCBY-NC-ND4.0) ... 13

FIGURE 4.PEMODEL FOR SOFTWARE EVOLUTION PRODUCTIVITY BY (WAGNER & DEISSENBOECK,2019)(CCBY-NC-ND4.0) ... 14

F^IGURE 5.CASE STUDY RESEARCH DESIGN PROCESS (RUNESON ET AL.,2012) ... 20

FIGURE 6.DESCRIPTION OVER CATEGORIZATION, FROM PROJECT RESOURCES TO SOURCING STRATEGY. ... 22

FIGURE 7.VISUALIZATION OVER THE SAMPLE SIZE PER PROJECT ARRANGEMENT. ... 25

FIGURE 8.OVERVIEW OF THE 64 PROJECTS PERFORMANCE MEASURES... 33

FIGURE 9.BOXPLOT OF PRODUCTIVITY AND SOURCING STRATEGIES. ... 34

FIGURE 10.BOXPLOT OF DEFECT DENSITY AND SOURCING STRATEGIES. ... 35

ix

A BBREVIATIONS

R&D Research and Development

GSE Global Software Engineering

GSD Global Software Development

ICT Information, Technology and Communication

x

T ERMINOLOGY

Defect An imperfection or deficiency in a project component

where that component does not meet its requirements or specifications and needs to be either repaired or replaced.

Defect density Number of defects per unit of product size (high defect density equals low quality)

Closed source A proprietary software used under contractual agreements.

Insourcing Leveraging company-internal resources

Offshoring Leveraging resources from a different country

Onshoring Leveraging resources from the same country

Open source A public software open to use, read or modify Outsourcing Leveraging external third-party resources

Personnel imbalance Extent of unevenness in the distribution of personnel across locations

Sourcing Leveraging resources

Temporal distance Time differences between countries

xi

[This page has been intentionally left blank]

1

1 I NTRODUCTION

Over the past three decades, increased connectivity, digitalization, and communication have simplified the use of sourcing strategies for firms of different sizes in many industries. This has facilitated the decision of going global and outsourcing core business activities and services (Rilla & Squicciarini, 2011; Khurana, 2006). The concept of outsourcing started in the 1950s but did not gain momentum and interest until what started as the first wave of domestic outsourcing of non-core manufacturing activities in the early 1980s (Hätönen & Eriksson, 2009).

During the first wave of outsourcing, the primary objective for firms was to cut costs. A decade later, the second wave of outsourcing followed. The concept evolved into strategic outsourcing which firms engaged in to enhance their capabilities and processes. Firms started to form strategic partnerships with vendors internationally to whom they outsourced activities they lacked expertise and competence in (Hätönen & Eriksson, 2009). Many labour-intensive activities such as IT and different types of customer services were outsourced (Metters &

Verma, 2008). The third and current wave of outsourcing, which started in the early 2000s, evolved into outsourcing of value-added and knowledge-intensive activities globally, such as R&D units which are of strategic importance to firms (Hätönen & Eriksson, 2009; Maskell et al., 2007). As a result, the current wave of outsourcing has led to the creation of the “barrier- less” organization (Hätönen & Eriksson, 2009).

The global technology race has become evident in many industries as a result of the “barrier- less” organization, which is reflected by high R&D expenditures in technology-intensive industries (Ebert, 2014, Hernández et al., 2019). Over the past 10 years, global R&D investments have grown by 67% and have been mainly concentrated on three major industries:

health, automotive and ICT (information, technology, communication). In 2018, the industries accounted for 76.6 % (ICT 38.7%, health 20.7% and automotive 17.2%) of the top 2500 global companies that invested more than €30 million in R&D activities. Notably, the rapid growth in R&D globally has to a large extent been driven by ICT producer and service sectors and the health industry for the past 15 years (Hernández et al., 2019).

The large increase in R&D for ICT producer and service sectors can be traced to the increased interest on global software development (GSD) which started 20 years ago (Herbsleb & Moitra, 2001; Ebert, 2014). GSD is closely coupled with outsourcing and relocation of core business and service activities. The concept stems from contract programming which has its roots in the second wave of outsourcing (Šmite et al., 2010; Lee et al., 2000; Hätönen & Eriksson, 2009).

GSD is conducted in geographically, temporally or culturally distributed projects with the objectives of developing software faster, cheaper and of higher quality (Šmite et al., 2010;

Herbsleb & Moitra, 2001). The objectives often result in different sourcing strategies which influence the way projects are conducted since it forces cooperation across firm or nation borders (Šmite et al., 2010). The project arrangements in GSD are mainly characterized by the following concepts (Barney et al., 2011; Šmite et al., 2014):

• Insourcing/Outsourcing – Defines whether the work is conducted by leveraging company internal resources or third-party consultants.

• Onshoring/Offshoring – Defines whether the work is conducted at the firm’s home country or in a different country.

2

The software industry is extremely susceptible to skilled competence due to the complexity and interdependence of the products’ (Conchúir et al., 2009; Sahay et al., 2003). Firms need to supply software development projects with the right competence and technology to sustain a competitive level since software work is knowledge-intensive and ever-changing (Sahay, et al., 2003; Ebert, 2016). As a result, the industry has become highly susceptible to various types of governance structures that include a bundle of sourcing strategies such as outsourcing and offshoring (Šmite et al., 2014; Ebert, 2016). Firm-specific objectives are often the foundation of a unique combination of sourcing strategies and thereby a firm’s governance structures can take different forms.

The choice of sourcing strategy for firms in the software industry is highly influenced by a financial factor; many firms face high costs which ultimately leads to a decision whether to make-or-buy a service i.e. develop it in-house or buy it from an IT vendor (Conchúir et al., 2009; Šmite et al., 2010). The decision is supported by two economic theories, the transaction cost theory and the resource-based view. Transaction cost theory takes the transaction costs of the activity into account when deciding on a sourcing strategy (Williamson, 1986). The resource-based view argues that firms' primary focus shall be on core business, keep core competencies within the firm and outsource the parts they lack competence (Prahalad & Hamel, 1990; McIvor, 2008). The two theories are highly important to consider since one of the objectives for firms to engage in offshoring activities is to cut costs. However, offshoring does not necessarily indicate a short-term economic gain (Šmite et al., 2017). Although a cost reduction of 90% in salaries can be enticing, not considering all transactions costs of offshoring can outweigh these benefits. In fact, relocating complex software development to an offshore IT-vendor may imply the activity to not reach break-even until after five years (Šmite et al., 2016). Hidden costs and low-performance evolution due to miscommunication, onboarding of new developers, and learning is associated with offshoring (Šmite & van Solingen, 2016; Britto et al., 2017).

Aside from cutting costs, another influential factor for firms in the software industry to relocate and outsource activities is the access to multi-skilled labour force which can be found in low- cost countries (Conchúir et al., 2009). The proximity to markets and customers are additional drivers for firms in their strategic decisions to engage in GSD (Conchúir et al., 2009). Similarly, customized technological infrastructures have attracted software multinationals of going global and outsource activities. For instance, technological parks with state-of-the-art telecommunication infrastructures were created in India for the sole purpose of attracting outsourced business activities to local IT vendors (Rao, 2004). India has been the top target country for outsourcing activities over three decades (Rao & Balasubrahmanya, 2017). The country remains at the top of outsourcing IT service destinations for multinationals according to the annual report of Kearney’s Global Services Location Index 2019 (GSLI). The index rates the countries based on financial attractiveness, people skills and availability, business environment, and digital resonance, places India at the top (Sethi et al., 2019).

Despite being early enticed by low-cost development, faster time-to-market, beneficial societal and legal structures, many firms have started to realize the challenges associated with GSD (e.g.

communication and coordination issues, quality issues, project failures and higher development costs) (Šmite et al., 2010; Conchúir et al., 2009). Firms have also started to realize the issues associated with sourcing strategies. For instance, outsourcing offshoring projects shown to be more prone to higher staff turnovers compared to insourcing projects which have shown to harm the project productivity and product quality (Bass et al., 2018).

3

1.1 P

Firms are striving to follow the value maximization principle which necessitates the need to actively reflect over the governance structure of the firm and decide on objectives. Individual organizations need to concretize and measure the performance of their actions, for instance by measuring what is better and worse. Accordingly, the best way forward which maximizes the long-term market value of the firm should be selected (Jensen, 2010). For software firms, the main concept related to value maximization and economic gain is software development. The development activities performed in projects are therefore highly relevant for firms to keep track of (Thamhain, 2003).

Developing software in a distributed setting in collaboration with third-party IT vendors is a relatively new phenomenon compared to the longstanding collaboration between manufacturers and external suppliers in the manufacturing industry. Yet, firms engaging in global work in other industries is not new since firms constantly strive to stay competitive and minimize costs (Sahay et al., 2003). Therefore, Sahay et. al (2003) emphasize the need to understand how sourcing strategies affect software projects in practice. For example, when GSD projects are distributed over nation and firm borders to maximize profits and competitiveness. Thus, understanding how the performance of the GSD projects, which already faces complexity in terms of high coordination costs and communications issues of multi-site collaborations, becomes more important. This since the addition of sourcing strategies in many projects adds to the complexity and risks already associated with GSD (Langer, 2016; Ebert, 2014).

The main reasons for firms’ engagement in GSD are to develop software products of high quality with short lifecycles and at low cost as mentioned earlier. However, the quality of the products often falls under scrutiny in a distributed development setting, even more so than when development is done in one site or by co-located teams (Carmel & Tija, 2005). Similarly, outsourced projects are evaluated by performance measures such as product quality and team efficiency where team efficiency can be considered a factor of productivity (Gopal & Gosain, 2010; Wagner & Deissenboeck, 2019). The project performance measures are needed to keep track of the increased complexity of communication and coordination that geographical and temporal distance bring in projects which eventually can result in higher development costs for the firms (Bird & Nagappan 2012; Herbsleb & Mockus 2003; Ramasubbu et al., 2011).

The existing literature is equivocal when it comes to the quality of the developed products in GSD projects. Some research reports that software quality decreases in distributed software projects (Bird & Nagappan, 2012; Ramasubbu & Balan 2007). Similar findings are reported for outsourcing projects, that is, small and medium-sized firms (SMEs) terminate their offshore outsourcing contracts due to low quality in the developed software as a result of poor external commitment (Moe et al., 2014). In contrast, existing research also reports little or no difference in quality when it comes to the software produced by co-located and distributed teams (Kocaguneli et al. 2013; Bird et al., 2009).

When firms engage in global work and outsource parts of the development to low-cost destinations, the process may also infer a negative impact on the software development productivity (Avritzer et al., 2010; Casey & Richardson, 2009; Milewski et al., 2008). Tsunoda et al. (2009) claim that projects with a high outsourcing level are less productive in comparison to projects with a low or medium level of outsourcing. Other researchers support the claim of distributed projects being less productive. For instance, when a project is offshored, the onshore teams spend more time on the prevalence of coordination and communication, and the

4

implication for the whole team is a decrease in productivity (Mishra & Mahanty, 2016).

Ramasubbu & Balan (2007) claim that this is due to the information asymmetry that arises in projects; coordination and task planning becomes more susceptible to misunderstandings which leads to a decrease in productivity. In contrast, other research (e.g. Ramasubbu et al., 2011) claims that productivity increases when firms distribute their software development projects across time zones, number of sites, personnel imbalance, and longer distances. Thus, previous literature is equivocal on how productivity is affected by global software development.

The performance of GSD projects, which can be measured by quality and productivity, accounts for the majority of the existing literature where the focus has been towards costs, geographical distance, distribution and performance evolution of new developers (e.g. Kocaguneli et al., 2013; Bird & Nagappan, 2012; Ramasubbu et al., 2011; Šmite et al., 2017; Britto et al., 2019).

However, single-sourcing strategies related to project performance has been studied to a lesser extent (e.g. Jabangwe et al., 2016; Tsunoda et al., 2009; Moe et al., 2014) while multiple and a combination of sourcing strategies (insourcing and outsourcing) related to project performance have been studied even less and remains unresearched. The business strategy of combining sourcing strategies is highly relevant for the “barrierless-organization” today as outsourcing of core business and service activities result in different strategic partnerships with service providers (Ebert, 2014; Langer, 2016). Thus, following the value maximization principle, firms need to evaluate project performance as a result of their business strategy.

To date and to the best of our knowledge, no research examines a combination of sourcing strategies (e.g. insourcing and outsourcing) and the relation to project performance in terms of quality and productivity. The reason is two-fold; first, leveraging internal employees and third- party consultants together in projects can take many forms (Langer et al., 2016). Second, a lack of assessing sourcing strategies in projects within a similar development setting where equal conditions apply is rare (Jabangwe et al., 2016).

5

1.2 P

The purpose of this thesis is to investigate how sourcing strategies relate to project performance in terms of quality and productivity. In particular, explore how the mixed strategy of insourcing and outsourcing performs in a project in relation to insourcing and outsourcing projects individually. To fulfil this purpose, a case study at a large-scale software firm that engages in distributed development will be conducted. Different project combinations of sourcing strategies can be explored in a real-life setting which enables an empirical investigation. It also enables to investigate the contribution of each sourcing strategy on the performance measures while controlling for the influence of other measures that affect project performance.

The contribution of this thesis will enrich and add to the existing literature on sourcing strategies and software project performance. Thus, we raise the following research question:

• What is the relation between sourcing strategies and software development project performance in terms of quality and productivity?

1.3 D

The scope of this study is delimited to investigate performance variations for different sourcing strategies within the ICT industry. A case firm engaging in distributed large-scale software development projects form the basis for the study. The study will be based on two sub-products at the case firm which share many similarities such as project arrangement, development processes, software metrics, and reasons for sourcing personnel from third-party vendors.

Performance has been limited to the project level and is measured in terms of quality and productivity where the metrics are anchored in existing theory. External factors affecting performance have been delimited to team size, location (onshore or offshore), cost and complexity of the projects. Thus, soft factors, organizational or cultural factors are not in the scope of this study. Further limitations in the result may have occurred from the chosen operationalization as quality and productivity are versatile concepts that can be operationalized in different ways.

Sourcing strategies have been limited to insourcing, outsourcing, co-located mixed sourcing (insourcing and outsourcing), and distributed mixed sourcing (insourcing and outsourcing) projects. The categorization of projects into the strategies are limited to “sourcing” and

“distribution”. Sourcing describes if the workforce is internal, external, or a mix of company internal employees and third-party vendor consultants. Distribution describes if the project is developed co-located or distributed across firms or nations.

6

1.4 O

An overview of the chapters in the thesis is presented as follows:

Chapter 2: Theoretical framework

This chapter presents a relevant theory about the approach of strategic sourcing and governance related to global software development and sourcing strategies.

Chapter 3: Related work

The related work is based on relevant papers, articles and books.

Chapter 4: Method

The method chapter describes the method selection and research design which includes research strategy, data collection, operationalization of the variables and models. The chapter ends with a section of threats to validity and ethical issues concerning the research.

Chapter 5: Result

In this chapter, results from the statistical tests are presented.

Chapter 6: Analysis

The results are analyzed for each performance measure and anchored in existing theory.

Chapter 7: Discussion

This chapter discusses the findings of the study and ends by highlighting the implications for research and practice.

Chapter 8: Conclusion

Chapter eight provides starts with a summary and conclusion of the research question. The chapter ends with a section describing future work.

7

2 T HEORETICAL F RAMEWORK

This chapter provides the reader with relevant theoretical background connected to the research question which forms the basis for this study.

2.1 S

Access to technology has changed the way firms decide on their sourcing strategies since it enables new communication possibilities. The accelerated way of communication has elevated possibilities for firms to source differently across their firm and nation borders. When deciding on a sourcing strategy, the main consideration is whether to make-or-buy the service (Embleton

& Wright, 1998). The make-or-buy decisions are critical for cost management in the majority of industries and it includes evaluating cost, profitability, efficiency, risk, and quality (Tayles

& Drury, 2001). Despite this, the primary objective for firms to outsource is to reduce cost, the reason is often to obtain short-term goals instead of considering what is most strategic from a long-term perspective (McIvor, 2000). The later wave of outsourcing including knowledge- intensive activities such as R&D has stressed the decision to find a more balanced strategy between the short- and long-term objectives to increase the probability to sustain on the market.

This decision has its roots in the theories of transaction cost and the resource-based view which will be presented in the next sub-sections.

2.1.1 T

The transaction cost theory is dominant in the existing literature explaining strategic sourcing decisions (McIvor, 2000; Hartman et al., 2017). The transaction cost theory aims to provide incentives for when activities are better to be performed within (insourcing) or outside (outsourcing) the firm and motivates when specific governance structures are best of being used. To make an informed decision, it shall include more than just increased capacity or decreased cost. The transaction cost theory enhances the possibilities of being economically efficient by maximizing performance meanwhile minimizing costs by taking an evaluation of resources and optimizing effort into account (Williamson, 1986). The framework of transaction theory does not only address the production costs, but also the transaction cost of, for instance, sourcing strategies. More specifically, when determining among various sourcing strategies, the theory enables assessing e.g. material-, capital- and labour cost with their corresponding transaction cost, e.g. coordinating, controlling, and monitoring (Lacity & Willcocks, 1996).

Williamson (1986) also argued that when the transaction cost of a good or service is high, the most appropriate governance structure is a hierarchy, and if the transaction costs are low, the good or service will be better of buying from the market, also mentioned “make-or-buy”.

Assessing transaction costs in practice can be challenging, but some example of conditions that aid insourcing more than outsourcing is; when few suppliers of the transaction are available, the specific transactions will occur, but also when transactions are asset-specific and faces high uncertainty (Lacity & Willcocks, 1996)

8

2.1.2 R

-

The necessity for firms to keep their main attention to their core business has increased since the rapid development of markets. The resource-based view argues that those activities which involve core competencies should be retained in-house (Prahalad & Hamel, 1990). Instead, processes and capabilities which the firm lacks competence in should be outsourced to gain a competitive advantage and to develop a superior performance over its competitors in the long- term (McIvor, 2008; Barney, 1991).

When a heterogeneity of knowledge-based resources exists in an organization, this can according to the theory of resource-based view explain differences in performance and competitive advantages (Teece et al., 1997). Thus, outsourcing aids the firm to attain a sustainable competitive advantage and superior performance by forming relationships with external providers and enhancing capabilities and processes.

2.2 G

Today, IT systems, services and software products are often developed in a global context. Due to the increased technological and communication capabilities, software products are nowadays a result of a multinational and complex supply chain developed from engineering management methods. Developing software in a global setting will evolve towards becoming a standardized method to use by every R&D manager. Over the past decade, globalizing software, IT- outsourcing, and business process outsourcing have faced a growth rate of 10-20% per annum.

However, working in a globally distributed context has its advantages and drawbacks (Ebert et al., 2016).

2.2.1 D

Incentives that drive firms to move their development activities to BRIC countries (Brazil, Russia, India, China) is the cheap labour cost and the large existing supply of educated personnel (Langer, 2016; Rao, 2004). Firms in numerous Asian countries with similar competences as in Europe and the USA charge a labour cost rate of 10-40% compared to onshore development (Ebert et al., 2016). Figure 1 presents the main drivers for firms to globalize and outsource their development processes based on experiences from the past 10 years of research (Ebert al., 2016).

9

Figure 1. A summary of drivers for GSD reported from firms in Europe and North America (Ebert et al., 2016).

Intrinsically, labour cost is the major reason for engaging in GSD since the possibility of decreasing the total project costs. However, the trend of outsourcing with regards to labour cost has shown to decline over time (Ebert et al., 2016). Additional advantages of GSD, besides the main driver of reducing costs, are the unique talent and skills, the reduced time-to-market (cycle time), flexibility and the proximity to local markets (see Figure 1). Other drivers are the access to unique technological infrastructure, the legal benefits such as patents and intellectual property but also the possibility of following the sun paradigm which enables firms to be operative 24 hours a day (Conchúir et al., 2009; Langer, 2016; Ebert et al., 2016).

Global software development poses challenges for the governance structure as the complexity of projects increase (Ebert, 2014). Coordination and communication efforts are growing and becoming more complex as the distribution across countries, firms and time-zones are increasing, regardless of that many IT vendors use sophisticated systems to cope with these issues and enable transparency (Sahay et al., 2003; Ebert, 2014).

GSD projects are exposed to a higher risk than on-site projects and more often lead to delays or failures (Ebert et al., 2001). Working in a globally distributed context can result in increased project cycle time, the effort required, and complexity due to more complex coordination and administration. A project with comparable complexity and size, developed in a cross-site collaboration takes longer to finalize in comparison to a project developed on-site. Such delays or failures have an impact on project performance (Ramasubbu & Balan 2007).

Colomo-Palacios et al. (2014) elaborate on other challenges that originate from global software development. Challenges with communication, coordination, control, efficiency, trust and sociocultural distance will be a part of the work environment for the employees in distributed projects. The challenges are often a result of the geographical, temporal, cultural and linguistical distance enhanced from distributing the development (Noll et al., 2010). Examples of challenges associated with such distances are, activities may occur at an inconvenient time for certain project members due to the time differences (Jarvenpaa & Leidner, 1999; Martins et al., 2004) or not being able to gather the members when receiving requirements or releasing products (Langer, 2016). Other examples of challenges are the differences in work moral, working hours and how employees can address the manager. Such differences may influence the trust among team members which are triggered by the sociocultural distance (Martins et al., 2004; Purna et al., 2011). Efficient knowledge sharing is considerably important for high-

32%

27%

21%

14%

6%

Labor cost Talent and skills Quality, cycle time Flexibility

Local market

10

performing teams, both distributed and co-located. Knowledge sharing can be performed written or verbally through direct contact (Gupta et al., 2009). Linguistic distance often results in various written and verbal communication skills which may influence the challenge of efficiency (Colomo-Palacios et al., 2014; Langer, 2016).

Many firms struggle to manage their distributed software development due to the encouragement of only seeing decreased cost and not evaluating the risks associated with the increased complex project organization. For instance, the cost of the initial learning curve increases by 20-40%, while within the first year 20% of sourcing contracts are cancelled, 50%

of all projects fails to deliver and 80% of firms are not satisfied with GSD activities (Ebert et al., 2016).

2.3 S

2.3.1 T

In line with the rapid growth in the IT industry, many firms have adopted new combinations of sourcing strategies that fit their firm-specific business model. As a result, sourcing strategies have become widely used as synonyms to describe a firm’s business model and the situation it is operating in. Therefore, a standardized empirically-based sourcing taxonomy for the software industry has been developed. This to reduce misunderstandings caused by the use of different terms that aims to describe similar cases, and to ease the challenges when searching for related topics in the literature (Šmite et al., 2014).

The taxonomy in Table 1 states that outsourcing refers to the leveraging of external third-party vendors whereas insourcing refers to the leveraging of company-internal human resources.

Offshoring and onshoring are distinguished by the location of the provided resources; onshoring refers to leveraging of resources in the same country and offshoring to leveraging of resources in a different country. Table 1 displays the different combination of sourcing strategies.

Table 1. Taxonomy for sourcing strategies (Šmite et al., 2014).

Term Description

Global Software Engineering Development of a software artefact across more than one location Insourcing Leveraging company-internal human resources

Nearshoring Leveraging resources from a neighbouring country

Offshore insourcing Leveraging company-internal resources situated in a different country Offshore outsourcing Leveraging external third-party resources situated in a different country Offshoring Leveraging resources from a different country

Onshore insourcing Leveraging company-internal resources situated in the same country Onshore outsourcing Leveraging external third-party resources situated in the same country Onshoring Leveraging resources from the same country

Outsourcing Leveraging external third-party resources

Sourcing Leveraging resources

11

2.3.2 P

In GSD, many different project arrangements include combinations of sourcing strategies and distributed development. However, the two do not imply the existence of the other, meaning distributed development can exist without a sourcing strategy, and vice versa (Šmite, 2014).

The project arrangements in GSD are many, for instance, distributed teams can operate locally in the physical facilities of a firm, be distributed nationally within one country, or distributed internationally among multiple nations (Griffith et al., 2003). Another possible outcome is that third-party vendors can assign personnel to projects conducted at the premises of the outsourced firm. In such arrangements, assigned personnel take part in projects on-site and act as if they were employed in that firm (see Figure 2, scenario 2) (Langer, 2016).

Figure 2. Project arrangements in GSD categorized by sourcing strategies, adapted from Šmite (2014).

1. Co-located team with internal employees 2. Co-located team with on-site consultants

3. Co-located outsourcing project 4. Non-distributed insourcing project

5. Distributed insourcing project 6. Distributed outsourcing project

7. Distributed outsourcing project with a

virtual team 8. Distributed insourcing project with a virtual team

Project member Project member from a third-party vendor

Project manager / Team lead Project manager / Team lead from a third-party vendor

Team boundaries Location boundaries

12

In Figure 2, eight project arrangements are illustrated, 1-4 are non-distributed software development projects and 5-8 are distributed software development projects. The following descriptions of the project arrangements are illustrated in Figure 2, (Šmite, 2014):

1. Represents a project where a co-located team works on-site.

2. Represents a project where consultants from a third-party vendor work together with internal employees on-site.

3. Represents an outsourcing project where the firm acquires software development from the third-party vendor. The team consists of consultants and is managed by an internal project manager/team lead.

4. Represent an insourcing project conducted by a co-located team which is governed by an internal project manager/team lead at a different site. All team members are internal employees.

5. Represents an insourcing project conducted by two separate and co-located teams at different sites. All team members are internal employees.

6. Represents an outsourcing project conducted by two separate and co-located teams at different sites. One team consisting of internal employees and the other of consultants from a third-party vendor.

7. Represents an outsourcing project that consists of a virtual team of internal employees and third-party vendor consultants.

8. Represents an insourcing project that consists of a virtual team from two sites of the firm.

The project arrangements shown in Figure 2 are only examples of how a project can be arranged in practice. In an industrial context, the structure and the number of teams working in a project can vary as well as the multiple combinations of insourcing and outsourcing, and onshoring and offshoring (Šmite, 2014).

2.4 P

,

Productivity is a versatile term and can be defined in many ways. It has evolved historically over the years and how to measure it varies between industries. A consensus among existing measures is that productivity describes a ratio between inputs and outputs. However, the way of measuring the units produced, over what time frame, is not straightforward (Wagner &

Deissenboeck, 2019).

Terms that are frequently used in association with productivity are efficiency, effectiveness and performance (Wagner & Deissenboeck, 2019). Efficiency and effectiveness are often mixed up and confused with productivity. Efficiency is an expression for “doing things right” and explains the application of resources. In the productivity ratio, it is formulated as “input”, see Figure 3. Effectiveness can be considered an expression for “doing the right things” or as an explanation for “usefulness and appropriateness”, also considered “output” in the productivity ratio, see Figure 3 (Wagner & Deissenboeck, 2019).

13

Quality is another aspect that is considered important for evaluating productivity since it encapsulates doing a good job meanwhile being productive (Wagner & Deissenboeck, 2019).

In a project, quality can be an internal consideration e.g. code quality, or an external consideration e.g. product quality from the end-user perspective.

Performance is considered a broader perspective and often covers a plethora of different factors that can influence a firms’ or a projects’ success (Wagner & Deissenboeck, 2019). Alexander (1994) claims that performance plays a key role when organizations opt to extend or change their sourcing strategy. Different well-established performance measures (e.g. balanced scorecard) include productivity in the formula but not exclusively or unique. A performance measure often involves other relevant factors to consider how successful the unit of measure is (Wagner & Deissenboeck, 2019).

2.4.1 P

Software development project performance can be measured in many ways but are often measured using the two metrics product quality and productivity (Krishnan et al., 2000; Wallace et al., 2004). Wallace et al. (2004) measure project performance as a result of product performance and process performance. The former refers to the “successfulness” of the product which can be measured by its quality. The latter refers to factors in the development process such as schedule and cost overruns which is related to the project’s efficiency and productivity.

Another way of describing performance is presented in Figure 4. The figure is illustrating how purpose, functionality, quality and effort relates to effectiveness, productivity, profitability, and performance.

Figure 3. Purposed-based effectiveness and efficiency by (Wagner & Deissenboeck, 2019) (CC BY-NC-ND 4.0)

14

Ramasubbu et al. (2011) project performance model stems from the economic view of measuring software development, i.e. from productivity and quality factors. Productivity can be measured as a ratio between output and input of a project, that is KLOC (Kilo Lines of Code) divided by project effort measured in person-hours. Quality can be measured as a ratio of KLOC and the number of defects reported from the customer.

2.5 T

• Transaction cost theory and resource-based view are two fundamental economic aspects that should be considered when firms opt to change their sourcing strategy to gain competitive advantage.

• The main drivers for GSD are decreased labour cost, increased quality, more flexibility and proximity to markets.

• The main drawbacks and challenges from GSD relate to increased complexity, coordination, communication, control, trust, cycle time, and effort.

• Many project arrangements in software firms can be traced down to different combinations of external and internal employees operating in a co-located or distributed environment.

• Project performance can be considered a plethora of factors. In software projects, it is often measured by productivity and quality.

Figure 4. PE Model for software evolution productivityby (Wagner &

Deissenboeck, 2019) (CC BY-NC-ND 4.0)

15

3 R ELATED W ORK

The following section presents a review of related work in the area of GSD project performance, sourcing strategies and other relevant studies related to the field of this thesis.

There exist numerous research papers that have examined how distribution aspects such as geographical, cultural and temporal distance affect the performance of software projects. For instance, a study by Ramasubbu et al. (2011) investigated how distribution dimensions such as geographical, temporal and configurational distribution (number of sites, personnel imbalance, experience spread) impact the performance of globally distributed software projects. The research was based on 362 projects from four different firms located across 15 countries distributed across 5 continents. The project performance was measured by project productivity, quality and profits. Productivity was measured as a ratio of total project code size and person- hour and quality as a ratio of total project code size and defects delivered after release to the customer. The profits were calculated as the ratio of project cost and revenue. The results from geographical and temporal distance indicated improved productivity and decreased quality for the projects. The authors also reported that different configurations of distributed development lead to different trade-offs since productivity and quality moved in opposite ways. For instance, the number of sites involved increased productivity while decreased quality and profits. In contrast, the experience imbalance increased quality and profits while decreased productivity.

A similar study addressing the same topic was conducted by Bird & Nagappan (2012) which investigated two distributed software projects related to two well-known open-source products, Eclipse and Firefox. The study aimed to investigate how they were impacted by organizational and geographical distribution in terms of software quality measured by pre-release and post- release defects. The study revealed that more pre-release and post-release defects could be attributed to software projects with high geographical distance (Firefox) whereas fewer defects to software components that are organizationally distributed (Eclipse). However, the authors stressed the importance of further studying distributed development and how it differs from co- located development as this is phenomenon is constantly evolving.

Ramasubbu & Balan (2007) conducted a two-year field study to investigate the effect of dispersion on project performance by investigating a multi-site distributed development setting of 42 distributed software projects. The proxy for quality was measured by pre-release defects and productivity as a ratio of code size in function points and total development effort measured in person-hours. In contrast to previous research discussed earlier, the results revealed a significant negative effect on productivity and an indirect effect on quality. This was because the quality was strongly correlated with productivity. The researchers claimed that a reason for these results might depend on the information asymmetry between teams which become enhanced in a distributed environment due to geographical distance. For instance, coordination and task planning become more complex and difficult which the authors' claim affects project performance. Concludingly, the authors emphasized that companies need to consider productivity and quality losses when moving development to other locations and that quality losses can be mitigated through software quality processes.

The project management aspect of GSD projects has also received attention in previous research since the communication and coordination among the teams are a vital part of a project’s success. A study by Colomo-Palacios et al. (2014) presented a paper related to this topic. They investigated how project managers affect productivity and performance in GSD. The sample

16

investigated was based on four IT-firms including 16 projects, developed by both offshore and in-house teams in relation to the location of the project manager. The authors claimed that the difference between planned- and actual duration is greater for offshore teams than for in-house teams. Offshore teams require more project management related effort than in-house teams and accordingly the offshore teams also report lower productivity than the in-house teams. Offshore teams also presented a greater defect rate than in-house teams. The authors concluded that the performance of project managers is affected negatively by the management of offshore projects.

Kormeren & Parvianen (2007) study research reported similar findings which showed that a decrease in productivity of 50% is apparent for globally distributed teams compared to co- located teams. This was also reported in an earlier study by Herbsleb & Mockus (2003) which showed that delivering products takes 2,5 times longer in a globally distributed environment compared to a co-located environment.

Only a few works study sourcing strategies together with software project performance. For instance, a study by Jabangwe et al. (2016) investigated how offshore outsourcing affects software quality and its evolution when the number of companies involved increases. The study reported inconclusive results due to the absence of pre-release defects in their quality proxy.

However, the authors claimed that pre-released defects, which refer to integration defects of software before releasing to the customer, offer a better measure when teams have worked independently in projects. Moreover, they discussed how measuring quality in different ways can result in incorrect inferences. This because some studies measure pre-release and some post-release defects, and in different project contexts. As a result, they claimed that the research has become equivocal regarding the quality of distributed software projects and urges more empirical evidence in more similar contexts to bridge the findings.

A study by Moe et al. (2013) conducted a multiple case study of four medium-size software development firms that cancelled their offshore outsourcing contracts and turned back to insourcing projects. This enabled them to discuss reasons from the previous empirical research regarding the unsuccessfulness of the outsourcing strategy. An explanation of why the firms terminated their contracts with their third-party vendors and turn to insourcing was due to low software quality. Their interviews with the case firms concluded the decreased quality to be a result of failures to accumulate the human and social capital, e.g. domain knowledge, absence of commitment from external employees, cultural differences, communication barriers and high turnovers which had a considerable effect on the projects.

Another study related to sourcing strategies conducted by Tsunoda et al. (2019) investigated productivity impacts and the relationship to different attributes of a software project. The study measured productivity impact from outsourcing, industry sectors, programming languages and team sizes by studying 211 projects in Japan. Their findings revealed that a small team relates to higher productivity in comparison to larger team size. The results from the different industry sectors reported that productivity in manufacturing projects was the highest while productivity in the business projects was the lowest. The most productive outsourcing project was the projects sourced with the” lowest” outsourcing ratio, and when focusing on project duration, the long duration projects were least productive.

Bass et al. (2018) studied the factors affecting employee retention by comparing offshore insourcing and offshore outsourcing. The objective was to better understand the effect of high staff turnovers in GSD which have shown to affect productivity and quality. The study, which comprised of a multiple-case study method and employed a quantitative and qualitative approach, found that insourcing employees have better work-life balance and are more driven

17

by customer satisfaction achieved by developing software of high quality. In contrast, team members in the outsourcing setting were also driven by customer satisfaction, but towards the client, i.e. the outsourcing firm, which the external developers were expected to satisfy.

Customer satisfaction was achieved by high development productivity, which often could result in working long hours to compensate for poor productivity. Furthermore, the authors explained that the external developers in the offshore outsourcing setting did not feel that the developed work belongs to them. While in the insourcing projects, the developers considered the development artefacts belong to the team and thus nurtured the product more with regards to the quality.

The communication and coordination among third-party consultants and internal employees have received attention recently. Nesheim & Hunskaar (2015) empirical study focused on knowledge sharing between external consultants and internal employees in large-scale projects which shed light on the less-discussed challenges of outsourcing. The study used a quantitative approach using a survey to collect data from 117 respondents where 63 % belonged to the “out- group” - external consultants. The study revealed that internal employees tend to share their knowledge with their “in-group” than with their “out-group” and vice versa.

Another perspective of performance has been studied by Britto et al. (2019) who investigated the performance evolution of employee onboarding in software development projects. The study elaborated on the difficulties of gaining knowledge of complex and large systems which can result in periods of low performance when onboarding new employees. The data collected in their case study was based on large-scale distributed software projects. The analysis resulted in four aspects that appear related to low-performance evolution: how distant the mentors were in relation to the newcomer, the training approach, allocation of specific tasks in early stages and instability of the team.

Šmite & van Solingen (2016) and Šmite et al. (2017) have focused on the additional and hidden costs associated with offshoring. Regardless of firms’ objectives to relocate activities across borders to save costs, the economic benefits are still questionable. The cost of salaries can imply a reduction of up to 90%, but not taking transaction costs under consideration can be harmful according to Šmite & van Solingen (2016). The study focused on mapping the costs related to offshoring. The conducted research was based on a collaboration between a software firm in the Netherlands and an IT-vendor in India. The study concluded that it can take up to 5 years to reach break-even for the initial investments. Factors such as knowledge transfer, travel, miscommunication, and learning are fundamental cost drivers (Šmite & van Solingen, 2016).

Despite the salary cost differences between sites, Šmite et al. (2017) report in a different study that after two years the high-cost locations remain “cheaper”. The implication of the result can be explained by i) complex products do not often generate economic gain from a short-term perspective and as aforementioned, break-even is hard to achieve even after five years, ii) in the studied case the motivation of offshoring might have been initiated in other objectives than cost.

18

3.1 B

• Research has been conducted on sourcing strategies separately e.g. insourcing or outsourcing.

• Firms' objectives to relocate activities to an offshore country or outsource to a third- party vendor are often to reduce costs, increase expertise or increase market proximity.

• Quality often decreases when firms distribute their software development globally.

• Productivity often decreases when firms distribute their software development globally.

• Hidden costs, information asymmetry, geographical distance, coordination and communications barriers are often associated with sourcing strategies that include other firms or nations.

Sourcing strategies related to project performance in software development have not been studied to a great extent. The majority of studies related to the software industry have focused on how co-located and distributed projects affect project performance. A new approach is therefore of interest, which our thesis addresses by analyzing how a combination of insourcing and outsourcing relates to project performance in software development. That is, how a mix of company internal employees and third-party consultants affects the project performance when cooperating in a co-located or distributed environment. The thesis will also investigate insourcing and outsourcing separate thus adding to the scarce literature of sourcing strategies but also to put the new combined sourcing approach in perspective.