See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/321713611

'Local content' of corporate social responsibility in international construction
business
Conference Paper · September 2017

CITATIONS

READS

0

129

5 authors, including:
Meng Ye

Roger Flanagan

The University of Hong Kong

University of Reading

13 PUBLICATIONS 27 CITATIONS

75 PUBLICATIONS 996 CITATIONS

SEE PROFILE

SEE PROFILE

Fan Xue

Dongping Fang

The University of Hong Kong

Tsinghua University

48 PUBLICATIONS 194 CITATIONS

96 PUBLICATIONS 2,288 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

MSc project: Multi-Agent Systems --- A distributed optimization approach View project

PhD project: Machine Learning-based generative hyper-heuristics View project

All content following this page was uploaded by Fan Xue on 14 May 2018.
The user has requested enhancement of the downloaded file.

SEE PROFILE

‘LOCAL CONTENT’ OF CORPORATE SOCIAL
RESPONSIBILITY IN INTERNATIONAL
CONSTRUCTION
Meng Ye1, Weisheng Lu2, Fan Xue3, Roger Flanagan4 and Dongping Fang5
1,2&3

Department of Real Estate and Construction, Faculty of Architecture, The University of Hong Kong,
Pokfulam Road, Hong Kong

3

School of Construction Management and Economics, Whiteknights, P O Box 225, Reading, Berkshire,
RG6 6AY, UK

4

Department of Construction Management, Tsinghua University, Haidian District, Beijing100084, People's
Republic of China.
‘Local content’ is increasingly stipulated in international construction business. While it
is agreed that ‘local content’ is a part of corporate social responsibility (CSR) agenda,
what exactly constitutes a ‘local content’ remains controversial. The primary aim of this
study is to (a) map out the status quo of ‘local content’ patterns as portrayed by
International Construction Companies (ICCs), and (b) offer deeper insights on the concept
by relating it to CSR localisation. By applying text mining to ICCs’ CSR reports, it is
found that local contents implemented in various host regions are largely uniform. As for
the indicators, ‘materials’ (EN1) and ‘Product and service labelling’ (PR2) are the most
reported local content initiatives. Besides, ICCs also pay much attention to implementing
other local content aspects, such as diversity and equal opportunity (LA4), local
communities (SO1), and public policy (SO3). These research findings help deepen the
understandings of local content and CSR concept, which could help ICCs and public
project sponsors to better promote CSR in the international construction context. The
findings could also be used for further investigation of the causal factors (e.g. impacts of
institutional distance) leading to the different local contents.

Keywords: local content, corporate social responsibility, CSR, localisation

INTRODUCTION
‘Local content’ is introduced as a part of CSR agenda by international companies recently
(Tordo et al. 2013). In line with the social impact compensation argument (Warner
2011; Ado 2013), local content can be viewed as the reasonable benefit in return to local
communities and economics, which may have been negatively affected by the operations
of international companies. Local content is always regarded as one strategy for the
resource-rich economies to avoid the ‘resource curse’ and to build sustainable capacity
for the local while they join the irreversible economy globalisation. It is also frequently
stipulated in the international construction markets, especially in the tender conditions on
requirements of an engagement of local labour, equipment and materials, and local
subcontractor (Han et al. 2010). In the construction market of Nigeria, for example,
local content is emphasized by encouraging local companies to participate and improve
their performance in the long run (Babatunde and Low 2013).

1

megan828@connect.hku.hk

This paper was presented as a working paper at the ARCOM 2017 Conference, Cambridge,
UK, 4-6 September 2017. Please contact the authors before citing

'Local Content' of CSR
Although local content is increasingly stipulated in international construction projects,
particularly for those sponsored by international organisations (e.g. the World Bank, the
African Development Bank), what exactly constitutes the ‘local content’ is still unclear.
Local content is a complex and fast-evolving area as mentioned by a World Bank study
(Tordo et al., 2013). It is generally agreed that Grossman (1981) provided the first and
most important intuitive foundation for the small theoretical literature in the field of local
content.
In his study, ‘domestic content’ was used instead of ‘local content’, which was
demonstrated as the percentage of domestic value added or domestic components
embodied in a specified final product. Grossman’s study was followed by many
theoretical works in the economics literature (Munson and Rosenblatt 1997; Lahiri and
Ono 1998; Qiu and Tao 2001; Lahiri and Ono 2003; Veloso 2006). In these works, local
content is regarded as one kind of requirements, stipulating that firms producing a good
are to procure a certain proportion of intermediate inputs domestically (UNIDO 1986;
Belderbos and Sleuwaegen 1997; Gu and Yabuuchi 2003; Kuntze and Moerenhout 2012).
For resource-rich countries, local content has a wider understanding that it refers to value
addition for the industry or the country by using local staff, local materials, and local
services (Heum et al., 2003). Its major objectives were transfer of technology, provision
of local job opportunities, creation of backward and forward linkages, and increase in
ownership and control (Ado 2013).
Putting it in the CSR agenda, local content is related to social impact compensation
(Warner 2011; Ado 2013), which argues for the adoption of local content policies as a
way to influence job creation and value addition for the local communities. This
proposed several questions: (a) what is the exact interpretation of local content? (b) What
is the relationship between local content and CSR localisation? (c) What is concluded in
local content for International Construction Companies (ICCs) to implement? and (d)
how do they implement the local content initiatives? Richer interpretations of ‘local
content’ and CSR concept are desired by both ICCs and public project sponsors with a
view to better promote both concepts in the international construction business.
The primary aim of this study is to (a) deepen the understanding of the concept of ‘local
content’ by relating it to CSR standardization and CSR localisation, and (b) map out the
status quo of ‘local content’ patterns as portrayed by ICCs. The remainder of this paper is
structured into four sections. Following the introduction section, the two main constructs,
local content and CSR localisation are elaborated. The research methods section provides
a detailed description of the sample and research methods, wherein text mining of the
CSR reports disclosed by ICCs is at the core. The fourth section presents the analyses,
findings and discussion. Conclusion and limitations are presented in the last section.
Constructs Explanation
CSR Localisation
CSR can be understood as a management concept, whereby companies integrate social
and environmental concerns in their business operations and in their interaction with their
stakeholders (Communities 2011). When companies act in different parts of the world,
their social and environmental concerns may be different due to the various cultures,
social norms and value systems of the respective stakeholders (Bustamante 2011). This
means that CSR activities need to be adapted to the specific situation of the region
wherein the companies operate (Wang et al., 2016). CSR localisation is introduced to
correspond to that, which refers to companies in host countries have a considerable

371

Ye, Lu, Xue, Flanagan and Fang
degree of autonomy to develop CSR strategies that are responsive to the local context and
local stakeholders (Muller 2006). Localisation strategy is always under consideration
compared with standardization strategy due to the “integration-responsiveness grid”,
describing the forces for global integration on the one hand and the local responsiveness
on the other hand (Prahalad and Doz 1987). As a strategy, CSR localisation would be
shaped by three main factors including the issue in question, the character and distribution
of major stakeholder groups and the general organizational approach and culture of a
company (Bustamante 2011). As for the issues in question, Bustamante (2011) made a
matrix measured by two aspects: degree of gloablity and degree of culture-sensitivity. In
his opinion, those issues with low degree of gloablity and high degree of culturesensitivity would more likely conform to the strategy of CSR localisation.
Local Content
Local content is similar to CSR localisation since they both emphasize the adaptation to
the specific situation of the region where a company operates. However, local content
holds different meanings which incorporate more culture-specific expectations and local
value addition than CSR localisation. On the one hand, local content is the
implementation of CSR strategies. It cares more about the CSR issues, which require
high degree of culture-sensitivity. In other words, those issues are characterized by a
strong cultural grounding and which would require the adaptation to the local
environment (Bustamante 2011). Poverty caring, social security or local employee
training projects are examples for that. In this perspective, local content could be
different from CSR localisation; the latter is a corporate strategy considering not only
CSR issues but also corporate stakeholders, strategic approach as well as organizational
culture. On the other hand, local content requires local value addition on the perspective
of the industry or the host countries. For example, the Republic of Trinidad and Tobago
(2004) defines local content as “local value-added” in terms of ownership, control and
financing by citizens of Trinidad and Tobago. Paul (2013) suggested the definition that
local content is the input to activities directly involved in operations along value chain,
which are provided by nationals. Therefore, local content is regarded as local CSR in this
study which refers to CSR implemented in the operating countries caring about local
culture and local value addition.

RESEARCH METHODS
Samples
The sample of ICCs was determined by referencing to the Top 250 International
Contractors List in 2015 compiled by ENR, an US-based construction, building and
engineering-oriented magazine. Some of the ICCs are excluded due to the lack of CSR
reports/sustainability reports. More ICCs in the Global Reporting Initiatives (GRI) list
collected from the GRI’s Sustainability Disclosure Database were added into the sample
in addition to ENR lists. As a result, seventy-five ICCs are chosen as the sample for the
analysis. They are from twenty-five countries. Nine companies from North America,
two from Latin America, thirty-four from Europe, twenty-seven from Asia and Australia,
and three from South Africa. CSR/sustainability reports of the sample ICCs over the past
five years (2011-2015) were retrieved from their websites or GRI’s Sustainability
Disclosure Database. There are 270 reports collected for the analysis due to some
missing reports in certain years.

372

'Local Content' of CSR
GRI Guidelines as the Analytical Framework
An analytical framework is identified for the further analysis. It is based on the GRI
Construction and Real Estate Sector Supplement (CRESS) G4, which provides indicators
and disclosures that are important or unique to the construction and real estate sector (Lu
et al., 2015). There are 6 categories with 28 sub-categories of performance indicators in
total (See Table 1).
Table 1: The protocol for decoding CSR reports
Code

Categories

Sub-categories

EC

Economic

EC1-Economic Performance; EC2- Market Presence; EC3- Indirect Economic
Impacts

EN

Environment

EN1- Materials; EN2-Energy; EN3-Water; EN4-Biodiversity; EN5-Emissions;
EN6-Effluents and waste; EN7-Products and services; EN8-Transport; EN9Land degradation, contamination and remediation

LA

Labour
practices

LA1-Employment; LA2-Occupational health and safety; LA3-Training and
education; LA4-Diversity and equal opportunity; LA5-Equal remuneration for
women and men

HR

Human rights

HR1-Non-discrimination; HR2-Child labour; HR3-Forced or compulsory
labour; HR4- Security practices; HR5-Supplier human rights assessment

SO

Society

SO1-Local communities; SO2-Anti-corruption; SO3-Public policy; SO4- Anticompetition behaviour

PR

Product
Responsibility

PR1-Customer health and safety; PR2-Product and service labelling

*Source: GRI Construction & Real Estate Sector Supplement G4

Text Mining
The analytical framework as shown in Table 1 is developed as the protocol for decoding
CSR/sustainability reports. Given the large volume of texts to be analysed, an innovative
method of text mining, rather than human decoding only, is used for the analysis of
CSR/sustainability reports. Although humans interpret the specific wording of a
document in the much larger context of their background knowledge and experience
(Gabrilovich and Markovitch 2007), text mining with computer algorithms could deal
with large number of texts and present a quick view of the collected reports. The process
of text mining is presented in Fig.1.
Step 1 is to convert collected CSR reports in PDF format to MS excel files. Step 2 is to
extract the text in the MS excel files accordingly corresponded to country names around
the world. There are 209 countries and other territories (e.g. Hong Kong, Bermuda)
identified for this process. This step presents 5,956 descriptions of local content of an
ICC in a certain year in a specific host country. In Step 3, stop words, such as function
words (is, having, when, etc.), pronoun (it, you, etc.), are eliminated from the
descriptions. Remained words with actual meanings are called ‘report words’.
Step 4 uses Explicit Semantic Analysis (ESA) method to do the text mining, which is
proposed by Gabrilovich and Markovitch (2007). The idea underlying ESA is to
represent and compare texts (from single terms to entire documents) as vectors in a high
dimensional concept space (Gottron et al., 2011). In this step, each report word is
assumed to convert to a vector weighted by the inverted index with 28 dimensions based
on the protocol mentioned above. Firstly, key words for each sub-category are identified
according to the interpretations of sub-categories in the GRI CRESS G4. For example,

373

Ye, Lu, Xue, Flanagan and Fang
for sub-category EN1-materials, 10 key words are identified such as ‘renewable
materials’, ‘raw materials’, etc.

Fig. 1: The process of text mining

Weighted inverted index is built based on the key words for 28 sub-categories. Secondly,
the semantic similarity between each report word and key word could be calculated,
encoded into the general word corpus (i.e. Wikipedia), by dividing the frequencies of both
words appearing in the Wikipedia corpus by the frequencies of each word. Thirdly, each
report word could be expressed by a vector with 28 sub-categories, using the maximum
semantic similarity between report words and key words in each sub-category. The
vector for one description is the maximum of the vectors of its component report words,
since it is assumed that the descriptions should be largely related to the sub-categories
when report words show the high semantic similarity. These processes convert the
descriptions into semantic similarity weighted vectors with 28 dimensions. When the
score for a dimension is higher, the description is more likely to be classified into the
corresponding sub-category.
Measurement of Local Content Similarity
In this study, each description represents local content implemented in a host country.
The similarity of two implemented local contents is assessed by the cosine similarity
between the corresponding weighted vector representations. Cosine similarity analysis is
commonly used for measuring similarity between two vectors (Singhal 2001). Two
vectors with the same orientation have a cosine similarity of 1, which means the
maximum score of cosine similarity is 1. When the score of cosine similarity is
approaching 1, the two vectors are similar. It could be implied that local content
374

'Local Content' of CSR
implemented in host countries is similar when the score calculated by cosine similarity
analysis is high.

ANALYSES, FINDINGS AND DISCUSSION
This study focuses on the regional market level only. The ENR adopts a convention to
classify ICCs’ operations in nine regional markets, namely, North America, Latin
America, Caribbean Islands, Europe, the Middle East, Asia/Australia, North America,
Central and Southern Africa, and Antarctic/Arctic. Africa is regarded as one continent
and few contractors operating in the Antarctic/Arctic. As a result, seven regional markets
are adopted. By merging host countries into the seven regional markets, the focus is on
how local content of ICCs is practiced in the host regions instead of host countries. This
has been incorporated in calculating the averaging semantic similarity weighted vectors
of local content in host countries.
Local content similarities among regional markets
ICCs from three home regions, i.e. North America, Europe and Asia/Australia are
selected as groups for the analysis respectively. In each group, local content similarity for
each two of the seven regional markets could be calculated and presented in Fig. 2. Each
colour block represents the local content similarities between two regional markets.
When the colour block is darker, it shows the higher similarity of the two regional
markets.

Fig. 2: Local content similarities among regional markets

It is shown that local content implemented by ICCs from North America, Europe and
Asia/Australia in various host regional markets is similar with all the similarity scores
larger than 0.9, although there also exist some slight differences. For ICCs from North
America (see Fig.2 (1)), they implement quite similar local content initiatives in North
America, Asia/Australia, Caribbean Islands, Latin America, and Africa, with the
similarity scores larger than 0.97. Noteworthy, they do the most different local content
initiatives in the Middle East compared to other host regions, evident by the lighter colour
blocks for Middle East. It seems that ICCs from North America emphasize indicators
related to environment very much, i.e. materials (EN1), energy (EN2), water (EN3) and
biodiversity (EN4), when operating in their own region, but the situation differs in the
regional market of Middle East. ICCs from European countries show the differential
local content implementations in various host regional markets (see Fig.2 (2)). North
American market shows a big divergence of local content implementations from other
host regional markets, especially with the market of Caribbean Islands. It is because that
European ICCs implement local content very well in North American market, especially
375

Ye, Lu, Xue, Flanagan and Fang
in the aspects of biodiversity (EN4), supplier human rights assessment (HR5) and
employment (LA1). For example, a UK ICC, Carillion, requires 100% of contracts will
have a biodiversity action plan in place to manage restoration when acting in Canada.
While in the market of Caribbean Islands, European ICCs seem not to concentrate on
local content implementation. For ICCs from Asian/Australian countries (Fig.2 (3)), they
have a relatively high differential local content initiatives in Caribbean Islands evident by
the relatively light colour blocks in Fig.2(3), but this may be caused as ICCs from
Asia/Australian do limited projects there.
Fig.2 presents high degree of similarity for each ICC implementing local content in
regional markets. This may be caused by the data which are extracted from CSR reports.
CSR reporting is under the pressure provided by the increasingly uniformed guidelines.
For example, the global reporting initiative (GRI) sets out minimum requirements in key
areas of CSR and companies sign up to report in accordance with the requirements (Lu et
al., 2015). The ISO 26000 also defines good practice and establishes standards (ISO
2010). Another explanation appears that differences may be reduced when considering
the region level instead of the country level. Moreover, it must be taken into account that
whether home region has larger impacts on local content implementations than host
regional markets.
Specific Indicator Analysis
As mentioned above, extracted CSR initiatives are the ones specifically implemented in
the host countries, which are regarded as local content initiatives. Based on the results of
Step 4 in text mining, semantic similarity weighted vectors with 28 dimensions for each
year could be calculated by making an average ignoring the host regional markets. The
score for each dimension shows the relevance of local content initiatives to each subcategory. Fig.3 presents the local content initiatives for the years 2011-2015. It can be
shown in the figure that Materials (EN1) and Product and service labelling (PR2) are the
most reported local content initiatives according to the keywords identified. Based on the
keywords analysis, specific materials, such as cement, steel, and concrete, are mentioned
most. Renewable materials, raw materials, recycled materials and materials reused are
frequently mentioned as the local content initiatives. Besides, product information,
service information and safe use of service are included as the key words which attract
remarkable considerations. ICCs also pay much attention to implement local content on
other indicators, such as diversity and equal opportunity (LA4), local communities (SO1),
and public policy (SO3). In the case of key words of these indicators, gender equality and
age group attract much attention. Besides, local content related to local communities is
always implemented by ICCs, such as welfare, young and children, work councils, and
community healthcare projects.

CONCLUSION AND LIMITATIONS
This study reviewed the concept of local content, and tried to further enrich the concept
by putting it into the context of international construction. Local content in this study
refers to those CSR activities tailor-made in different host countries while conducting
international construction business. 270 CSR/sustainability reports of ICCs over the past
five years (2011-2015) are collected as the data source for analysing ICC’s local content
patterns. CSR initiatives are extracted from the CSR/sustainability reports in line with
their host country names. It is assumed that only those countries where ICCs implement
specific CSR initiatives would be mentioned or stressed in the reports. Therefore,
extracted CSR initiatives are the ones specifically implemented in the host countries,
which are regarded as local content initiatives.
376

'Local Content' of CSR
By triangulating the results from computer-aided text mining and human analyses, it is
found that local content implemented in various host regions is similar. This may be
caused by the joint, yet dilemmatic isomorphic forces of CSR (reporting) standardization
on the one hand while CSR localisation on the other. However, their local content
implementations may be affected by both home regions and host regions. As for the
indicators, Materials (EN1) and Product and service labelling (PR2) are the most reported
local content initiatives. Besides, Diversity and equal opportunity (LA4), Local
communities (SO1), and public policy (SO3) also attract much attention of ICCs.

Fig. 3: Local content initiatives implemented from 2011 to 2015

Further investigations on different levels are needed especially when they are placed in
the context of institutional theory. Furthermore, the method of text mining is largely
depended on the selection of key words, which would directly affect the results, therefore,
the process of text mining including the selection of key words needs to be deeply
considered.

REFERENCES
Ado, R (2013) Local content policy and the WTO rules of trade-related investment measures
(TRIMS): the pros and cons. International Journal of Business and Management Studies,
2(1), 137-146.
Babatunde, O K, and Low, S P (2013) Chinese construction firms in the Nigerian construction
industry. Habitat International, 40, 18-24.
Belderbos, R and Sleuwaegen, L (1997) Local content requirements and vertical market structure.
European Journal of Political Economy, 13(1), 101-119.
Bustamante, S (2011) Localization vs. Standardization: Global approaches to CSR Management
in multinational companies. Working Papers of the Institute of Management, Berlin:
Berlin School of Economics and Law (HWR Berlin).
Communities (2001) Green Paper: Promoting a European framework for Corporate Social
Responsibility. Brussels: Commission of the European Community.

377

Ye, Lu, Xue, Flanagan and Fang
Gabrilovich, E and Markovitch, S (2007) Computing semantic relatedness using Wikipedia-based
explicit semantic analysis. In: Proceedings of International Joint Conference on Artificial
Intelligence (IJCAI-7), 1606-1611.
Gottron, T, Anderka, M and Stein, B (2011) Insights into explicit semantic analysis. In:
Proceedings of the 20th ACM International Conference on Information and Knowledge
Management, ACM, 1961-1964.
Grossman, G M (1981) The theory of domestic content protection and content preference. The
Quarterly Journal of Economics, 96(4), 583-603.
Gu, W and Yabuuchi, S (2003) Local content requirements and urban unemployment.
International Review of Economics & Finance, 12(4), 481-494.
Han, S H, Kim, D Y, Jang, H S and Choi, S (2010) Strategies for contractors to sustain growth in
the global construction market. Habitat International, 34(1), 1-10.
Heum, P, Quale, C, Karlsen, J E, Kragha, M, and Osahon, G (2003) Enhancement of local content
in the upstream oil and gas industry in Nigeria: A comprehensive and viable policy
approach. SNF Report No. 25/03, Bergan, Norway: Institute for Research in Economics
and Business Administration.
ISO (2010) Social responsibility. ISO 26000. Geneva: International Standards Organisation.
Kuntze, J-C, and Moerenhout, T (2012) Local content requirements and the renewable energy
industry-a good match? Ad hoc Expert Group Meeting on Domestic Requirements and
Support Measures in Green Sectors: Economic and Environmental Effectiveness and
Implications for Trade, 13-14 June, Salle XXI, Palais des Nations, Geneva: International
Centre for Trade and Sustainable Development/ Global Green Growth Institute
Lahiri, S, and Ono, Y (1998) Foreign direct investment, local content requirement, and profit
taxation. The Economic Journal, 108(447), 444-457.
Lahiri, S and Ono, Y (2003) Export-Oriented foreign direct investment and local content
requirement. Pacific Economic Review, 8(1), 1-14.
Lu, W S, Ye, M, Flanagan, R and Ye, K H (2015) Corporate Social Responsibility Disclosures in
international construction business: Trends and prospects. Journal of Construction
Engineering and Management, 142(1), p.04015053.
Muller, A (2006) Global versus local CSR strategies. European Management Journal, 24(2), 189198.
Munson, C L and Rosenblatt, M J (1997) The impact of local content rules on global sourcing
decisions. Production and Operations Management, 6(3), 277-290.
Paul, A E (2013) What is Local Content? -An overview. In: Conference on Local Content
Policies in the Oil, Gas and Mining Sectors, 30th September-1st October, Vienna,
Austria: World Bank’s Sustainable Energy Department and Oil, Gas and Mining Unit.
Prahalad, C K, and Doz, Y L (1987) The Multinational Mission: Balancing Local Demands and
Global Vision. New York: The Free Press.
Qiu, L D, and Tao, Z (2001) Export, foreign direct investment, and local content requirement.
Journal of Development Economics, 66(1), 101-125.
Singhal, A (2001) Modern information retrieval: A brief overview. (IEEE) The Bulletin of the
Technical Committee on Data Engineering, 24(4), 35-43.
The Republic of Trinidad and Tobago (2004) Local Content and Local Participation Policy
Framework for the Republic of Trinidad and Tobago. Port of Spain, Trinidad and
Tobago: Ministry of Energy and Energy Industries.

378

'Local Content' of CSR
Tordo, S, Tracy, B S, and Arfaa, N (2011) National Oil Companies and Value Creation.
Washington DC: World Bank Publications.
Tordo, S, Warner, M, Manzano, O and Anouti, Y (2013) Local Content Policies in the Oil and
Gas Sector. Washington DC: World Bank Publications.
UNIDO (1986) Industrial policy in the developing countries: An analysis of local content
regulations. Vienna: United Nations Industrial Development Organization.
Veloso, F M (2006) Understanding local content decisions: Economic analysis and an application
to the automotive industry. Journal of Regional Science, 46(4), 747-772.
Wang, H D, Lu, W S, Ye, M, Chau, K W and Zhang, X L (2016) The curvilinear relationship
between corporate social performance and corporate financial performance: Evidence
from the international construction industry. Journal of Cleaner Production, 137(11),
1313-1322.
Warner, M (2011) Local Content in Procurement: Creating Local Jobs and Competitive Domestic
Industries in Supply Chains. Sheffield, UK: Greenleaf Publishing.

379

View publication stats