ITC WORKING PAPER SERIES - International Trade Centre€¦ · ITC WORKING PAPER SERIES EXPLORING...

TRADE IMPACTFOR GOOD

ITC WORKING PAPER SERIES

EXPLORING FIRM COMPETITIVENESS: A FACTOR ANALYSIS APPROACH

WP-04-2017.E

December 2017

Justine FalciolaUniversity of Geneva

Marion Jansen International Trade Centre

Valentina RolloInternational Trade Centre

Disclaimer

Views expressed in this paper are those of the authors and do not necessarily coincide with those of ITC, UN or WTO. The designations employed and the presentation of material in this paper do not imply the expression of any opinion whatsoever on the part of the International Trade Centre or the World Trade Organization concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. Mention of firms, products and product brands does not imply the endorsement of ITC or the WTO. This is a working paper, and hence it represents research in progress and is published to elicit comments and keep further debate.

© International Trade Centre WP-04-2017.E

ITC Working Paper Series

EXPLORING FIRM COMPETITIVENESS: A

FACTOR ANALYSIS APPROACH

December 2017

Justine Falciola

University of Geneva

Marion Jansen

International Trade Centre

Valentina Rollo

International Trade Centre

Disclaimer

Views expressed in this paper are those of the authors and do not necessarily coincide

with those of ITC, UN or WTO. The designations employed and the presentation of

material in this paper do not imply the expression of any opinion whatsoever on the

part of the International Trade Centre or the World Trade Organization concerning the

legal status of any country, territory, city or area or of its authorities, or concerning the

delimitation of its frontiers or boundaries. Mention of firms, products and product

brands does not imply the endorsement of ITC or the WTO. This is a working paper,

and hence it represents research in progress and is published to elicit comments and

keep further debate.

Exploring firm competitiveness: a factor analysis approach

Justine Falciola1 Marion Jansen2 Valentina Rollo3

UNIGE ITC ITC

Abstract

This paper uses confirmatory factor analysis (CFA) to build an index of firm competitiveness and fill

a gap in the literature. The proposed competitiveness framework and its subcomponents, tested

by the CFA, are identified according to the review of the economic and management literature and

related empirical evidence. We use data from the World Bank Enterprise Surveys for 100 countries

of different income and development status. Our results suggest that the competitiveness index is

positively correlated with commonly used proxies of competitiveness, such as labour productivity,

the probability to export, the percentage of inputs of foreign origin used by the firm and the share

of total sales that were exported. Moreover, the competitiveness framework proves to apply to

firms of different sizes and to both exporting and non-exporting firms.

Keywords: competitiveness, factor analysis, latent variable models, multi-dimensional index, firm heterogeneity

JEL classification: F23, C38, M21, L11

1 Justine Falciola, Doctoral Student, Geneva School of Economics and Management, University of Geneva,

Uni-Mail, 1221 Geneva 4, Switzerland; [email protected].

2 Marion Jansen, Chief Economist, Office of the Chief Economist, International Trade Centre, Palais des Nations,

1211 Geneva 10, Switzerland; e-mail: [email protected].

3 Corresponding author: Valentina Rollo, Economist, Office of the Chief Economist, International Trade Centre,

Palais des Nations, 1211 Geneva 10, Switzerland; tel. +41-22-730.0331 ; e-mail: [email protected].

The authors thank Jaya Krishnakumar, Stephan Sperlich, Virginie Trachel, Olga Solleder and the participants of the UNIGE BBL in Geneva and the 2016 ETSG Conference in Helsinki for useful comments and discussions. We thank Yuliya Burgunder for help with the review of the literature.

mailto:[email protected]:[email protected]:[email protected]

1. Introduction

This paper uses multilevel confirmatory factor analysis to build an index of firm

competitiveness across countries.4 This is relevant and necessary because productivity

remains to date the most commonly used indicator of good performance and

competitiveness, at both the macro and micro level. Most importantly, whether productivity

fully represents the performance or competitive strength of contemporary organizations

remains a subject of discussion: a consensus on what is a good definition of productivity is still

missing, since no definition really captures all aspects of production, especially its dynamic

nature.

Production includes both tangible and non-tangible assets, such as knowledge work and

services (Oeij et al, 2011). This has been taken into account in work where non-tangible assets

are included in the definition of productivity - including the time factor (Johnston and Jones,

2004), quality (Drucker, 1999; Grönroos and Ojasalo, 2004), the role of clients or customers

(Martin et al., 2001), value creation (Rutkauskas and Paulavičiene, 2005) and capacity planning

(McLaughlin and Coffey, 1990; Jääskeläinen and Lönnqvist, 2009).

A part from access to knowledge, firms also need to be able to absorb capacity (Cohen &

Levinthal, 1990; Kim, 1997). To this end, R&D (Griffith, Redding, and Van Reenen, 2004,

Fagerberg and Verspagen, 2002), education (or human capital) (Barro, 1991; Benhabib &

Spiegel, 1994), finance (King & Levine, 1993; Levine, 1997; Levine & Zervos, 1998), and

governance (Acemoglu, Johnson, & Robinson, 2001; Glaeser et al., 2004; Rodrik et al., 2004)

play an important role.

The result is a proliferation of combinations of variables to define productivity.

The concept of competitiveness is not new; it has been described in the economic and

business literature as a multidimensional concept, where different criteria of competitiveness

depend on time and context (Ambastha and Momaya, 2004). Porter (1998) states that “it is

the firms, not nations, which compete in international markets”. Empirical evidence shows

that 36 per cent of the variance in firms’ profitability should be attributed to the

characteristics and actions of firms (McGahan, 1999), while other works focus on firms’

strategies and resource positions (Bartlett and Ghoshal, 1989; Prahalad and Doz, and 1987;

Prahalad and Hamel, 1990) as the real sources of competitiveness.

The environmental factors, in this paper divided between the national and the business

ecosystem, remain relatively uniform across all competing firms, but are crucial to the

4 The use of composite indicators in economics and business is very commmon, especially in industrial competitiveness, sustainable development, quality of life assessment, globalisation, innovation or academic performance (see Cox et al 1992, Cribari-Neto et al 1999, Färe et al. 1994, Griliches 1990, Forni et al. 2001, Huggins 2003, Grupp and Mogee 2004, Lovell et al. 1995, Author 2005, Author et al. 2005, Saisana and Tarantola 2002, and Wilson and Jones 2002, among others).

competitiveness of the firm.5 In fact, competitiveness arises from an integral process that goes

beyond the boundaries of the single firm and connects employees and clients/customers in

many ways (Oeij et al, 2011).

The challenging task to be tackled is to summarize several dimensions into one single measure

of competitiveness that would allow policy makers to monitor progress efficiently. This paper

tries to achieve this objective by shaping this multi-dimensionality into an index of firm

competitiveness. Since competitiveness is a latent concept, we use a latent variable model.

The choice of confirmatory factor analysis (CFA) is motivated by the fact that, based on the

review of the literature and empirical evidence, we hypothesise a competitiveness framework

that CFA allows us to tests statistically.

CFA differs in spirit from classical regression analysis as it emphasizes covariances rather than

individual variables. In fact, while multivariate regression analysis focus on the relation

between one or more known independent variables, 𝑥𝑖, and the known dependent variable

𝑦𝑖, factor analysis focuses on uncovering and making use of the relationship (and consequently

correlation) among observed indicators (the independent variables) in order to measure a

latent concept: competitiveness. Since most of the indicators included in the CFA are highly

correlated, a multivariate regression analysis would suffer from multi-collinearity. On the

contrary, CFA explicitly make use of the high correlation between indicators and is therefore

particularly suited for the construction of our competitiveness index.

The results suggest that our competitiveness index is positively correlated with commonly

used proxies of competitiveness, such as labour productivity, the probability to export, the

percentage of inputs of foreign origin used by the firm and the share of total sales that were

exported. Finally, the competitiveness framework we build is applicable to firms of different

size and to both exporting and non-exporting firms, as shown by the positive relationship

between labour productivity and the index for the different types of firms.

The contribution of this paper is therefore twofold. On one side it contributes to filling a gap

in the attempt to measure competitiveness, until now mainly proxied with several and open

to discussion measures of productivity. It proposes to measure competitiveness by building a

composite indicator, using confirmatory factor analysis, so assuming that competitiveness is

a latent concept that is unknown a priory. On the other side, this paper provides a first attempt

to measure competitiveness with factor analysis at the firm level, and it does so by proposing

and testing a competitiveness framework, based on the review of the economic and

management literature.

The rest of the paper is structured as follows. Section 2 provides a review of the literature,

while Section 3 introduces the confirmatory factor analysis, including the competitiveness

5 McGahan (1999) argues that only 36 per cent of the variance in firms’ profitability should be attributed to the characteristics and actions of firms. This is also argued and shown by Bartlett and Ghoshal (1989), Prahalad and Doz (1987) and Prahalad and Hamel (1990).

framework to be tested and the data. Section 4 test the relevance of the index in regression

analysis and finally Section 5 concludes.

2. Review of the literature

a. Components of firm competitiveness A multitude of components can influence the ability of a firm to perform well. These

components (highlighted either in the economic or in the business literature) can be directly

related to the characteristics of the firm (its innovativeness, its export status, access to a bank

account, the ability of the manager, etc.), or indirectly affect the firm through its business

environment. The latest can be further separated into immediate and macroeconomic

environment, according to whether it is close to the firm (clients, suppliers, competitors, etc)

or further away (national infrastructure, governance, trade policy, etc) in terms of connection

and ability to influence.

Moreover, since firms do not only need to compete today, but rather need to stay competitive

over time, it is important to take into account not only the static but also the dynamic

components of competitiveness.6 “Productive efficiency” and “dynamic efficiency” are

increasingly highlighted by the theoretical and empirical literature on gains from competition.7

Research in evolutionary economics, behavioural theory of the firm and transaction costs

economics has led to formulate the concept of dynamic capabilities of firms, which are tightly

related to the ability of firms active in international markets to shape international

environment, thus influencing a nation institutional framework (Dunning and Lundan, 2010).

Firms operating in a global environment are constantly exposed to change, and adequate

returns can only be achieved in a sustained manner if the firm is able to adjust to, or to

embrace, change.8

Managerial competence

One of the important components of firm competitiveness and a good predictor of how well

a firm will perform in the market is the competency of its manager.9 The subject has been

extensively developed by management, institutional and organizational studies, since

Hambrick and Mason (1984) discussed the relevance of managerial characteristics for

organisational outcomes. Management practices can improve productivity, through their

impact on marginal productivity of inputs and resource constraints (e.g. Syverson, 2011), as

well as growth and longevity (Bloom and Van Reenen, 2010). Learning even elementary

management skills in planning, marketing and financial literacy can lead to an accelerated

6 Especially in more “dynamically competitive” industries (Bresnahan, 1998; Evans and Schmalensee, 2001; Ellig and Lin, 2001). 7 Spence (1984), Ahn, (2002), Feurer and Chaharbaghi (1994) 8 As Nelson (1996) reminded, Schumpeter’s idea from his Theory of Economic Development “Static analysis is not only unable to predict the consequences of discretionary changes in the traditional ways of doing things; it can neither explain the occurrence of such productive revolutions nor the phenomena which accompany them. It can only investigate the new equilibrium position after the changes have occurred”. 9 Porter (1990) defines entrepreneurial and management skills as the ability to capitalize on ideas and opportunities by successfully implementing a business strategy.

adoption of improved management practices, increased willingness of owners to pay for

follow-up training and increased survival (Sonobe and Otsuka, 2006, 2011). However, variance

in organisational outcomes may be better explained by managers’ characteristics when there

is a higher degree of managerial discretion (Hambrick and Finkelstein, 1987).10

Years of managers’ experience are found to affect performance as well. Empirical evidence

shows that managers from an older generation are: more conservative in terms of investment

choices and use of financial leverage; more likely to undertake diversification moves and R&D

activities; more associated with higher returns on assets. At the same time, having an MBA

degree is related to more aggressive strategies and is also positively associated with higher

firm performance (Bertrand and Schoar, 2003).

Managerial skills also influence the firm’s capacity to internationalize. The effort to learn

internationally, together with previously acquired international experience and an open-

minded attitude towards global markets, all positively relate to internationalization (De Clerq

et al., 2005, Reuber and Fischer, 1997, Kyvik et al., 2013), being it entry into exporting (Wood

et al., 2015), or the capacity to diversify geographically (Ciravegna et al., 2014). The structure

of ownership also influences the decision to internationalize. Fernandez and Nieto (2005)

show that family-owned firms (commonly but not exclusively SMEs) engage less in

commitment-intensive internationalization activities. However, when SMEs are managed by

a group of shareholders, which include foreign shareholders, export propensity increases.

Quality and sustainability standards

Standards, weather national or international, affect the basic functioning of the firm (ITC,

2016). Adopting standards may increase sales on foreign markets, improve the image of a

company, or even decrease associate trade costs due to facilitated custom control regime

(Masakure, Cranfield and Henson, 2011; Latouche and Chevassus-Lozza, 2015; Volpe

Martincus, Carballo and Graziano, 2015). However, compliance with resource demanding

standards can require additional investment and financing in order to adjust the production

process, product labelling, packaging, etc. Consequently, certification may restrain producers

in accessing foreign markets, since they incur in extra costs, both fixed and variable, which

ultimately increase the product price (World Bank, 2005; Kox and Nordås van Tongeren, 2007;

Beghin and Marette, 2009).

The exhaustive available literature on the effect of ISO 9001 standards shows that

management system standards (MSSs) have enjoyed enormous success over the last years. A

review of the literature by Heras, Molina-Azorín and Tarí (2012)11 shows that the positive

effect of ISO 9001 and ISO 14001 standards are related to: improved efficiency and

effectiveness of the organization; a reduction of bureaucracy; a reduction in the costs of

10 A review of the literature devoted to the studies analysing managerial discretion can be found in Wangrow, Schepker and Barker (2015). 11 http://upcommons.upc.edu/bitstream/handle/2099/12955/tari.pdf

http://upcommons.upc.edu/bitstream/handle/2099/12955/tari.pdf

internal and external audits; and the availability of joint training and improved communication

between all organizational levels.12

However, the decision and possibility to comply with national or international standards only

partly depends on the firm’s capacities. Compliance also depends on the infrastructure

available in a country, being it terms of access to finance or physical infrastructure, or being it

in terms of supportive local or national institutions to provide information and guidance.

Access to finance

At the macroeconomic level, evidence shows that financial development matters for output

growth of the economy (Levine and Zervos, 1998), as it also affects growth potential of credit-

constrained firms (Rajan and Zingales, 1998). In order to operate, firms need to have a bank-

account to settle accounts with clients and providers quickly and smoothly. Investment in new

activities further requires access to finance. Musso and Schiavo (2008) show how access to

external finance in France has a positive effect on firm performance in terms of sales, capital

stock and employment. Access to finance is consistently cited as one of the primary obstacles

affecting SMEs more than large firms (Ayyagari, Demirgüç-Kunt and Maksimovic, 2012).

Access to finance is proved to be an important determinant of firm performance along a number of distinct aspects, including investment, growth, firm size distribution (Ayyagari, Demirgüç-Kunt; and Maksimovic, 2011), and innovation (Demirgüç-Kunt, Beck, and Honohan 2008). It also determines the firm’s ability to enter export markets and expand abroad (Bellone et al., 2010; and Berman and Héricourt, 2010), which are capital intensive efforts, involving high up-front costs (i.e. needed to create distributor networks) and high variable costs (related to shipping, logistics and trade compliance).

However, firms’ abilities and capacities are not the only element determining access to finance. The access to and extension of credit greatly depends on a supportive legal and regulatory framework. Coricelli et al. (2010) shows that in countries characterized by weak financial market institutions and limited market capitalization, a significant proportion of firms have no access to bank loans.

Access to talent

A skilled workforce is central to the ability of firms to anticipate change or to adjust to it, and

an important determinant of economic growth (Woessman, 2011). Backman (2014) provides

evidence of the link between work force education, experience and cognitive skills and firm

productivity. Local availability of talented workforce is not only a strong predictor of

productivity, but also of export diversification (Cadot, Carrère and Strauss-Kahn, 2011).

Matching the skills needs firms have with the skills supplied by countries’ education systems

is not always an easy task, and a usual source of inefficiency (Jansen and Lanz, 2013).

Talent is even more important in developing countries, where firms need

absorptive capacities to internalize foreign technologies, and where workers with education

and training are in high needs for this task. Firms that adopt new foreign technologies need

12 Wilkinson & Dale, 1999a, 1999b; Poksinska et al. 2003; Zeng, Tian & Shi, 2005; Zutshi & Sohal, 2005

educated staff to innovate as they enter more knowledge-intensive activities. This is even

more relevant when firms want to enter Global Value Chains. Evidence shows that firms in

countries with relatively low (high) skill levels receive low (high) skill-intensive tasks (Khalifa

and Mengova, 2012).

Some firms, especially SMEs, might need to invest in training but do not, simply because the

expected rate of return associated with training is smaller than the return on other

investments (Almeida, Behrman and Robalino, 2012). This is related to SMEs being more

resource constrained than larger firms (Okada, 2004), and to the difficulty for very small firms

to handle the drop in production that results from the absence of an employee in formal

training. The business ecosystem and the national environment can strengthen the

engagement of SMEs in training through cooperation via horizontal networks. These networks

can in fact create opportunities for knowledge exchange, resulting in collaborative research

and development (Bosworth and Stanfield, 2009).

Access to inputs and customers

In order to produce their final goods, firms need access to a varied range of inputs and

suppliers, and in order to sell, firms need access to customers (access to market). Empirical

evidence shows that access to foreign intermediate inputs can increase firms' efficiency by

providing more diverse and higher quality inputs (Bas and Strauss-Kahn, 2014), especially for

SMEs, since they are able to raise their productivity via learning, variety and quality effects

(Amiti and Konings, 2007). Importing also improves firm productivity (Vogel and Wagner, 2010

and Kasahara and Rodrigue, 2008). As a consequence importing can have a positive effect on

the decision to start exporting and also on the variety of products exported and success as an

exporter (Kasahara and Lapham, 2006; Bas and Strauss-Kahn, 2014).13

Even though the decision to export depends on the firm, access to market remains outside of

the firm’s control, as it is determined by the trade policy of home or destination countries.

Ample evidence shows that trade liberalization - lower tariffs and fewer barriers to trade -

leads to better economic outcomes (Wacziarq and Welch, 2008). Amiti and Konings (2007)

even show that reducing input tariffs increases productivity three times more than a reduction

in output tariffs. Trade liberalization does not only affect the capacity of a single firm to export

or import, it affects the degree of competitiveness firms face in a market (Melitz, 2003; Melitz

and Ottaviano, 2008).

Firms’ ability to import or export might also be constrained by logistics. Poor logistics

management can render firms uncompetitive, impeding their access to suppliers and buyers,

and their participation in international value chains. Logistics costs are an important share of

the value of final goods produced, especially for SMEs, and in developing countries: for

example, in LAC logistics costs represent 18% to 35% of the final value of goods, while in OECD

13 The trade literature has vastly proved exporting firm to be larger, more productive, more capital-intensive, more technology-intensive and pay higher wages than non-exporting firms (Bernard, A. B., & Jensen, J. B., 1999; Delgado, Farinas and Ruano, 2002).

countries it remains close to 8%. For small companies, the share may be over 42%, mainly due

to high inventory and warehousing costs (Schwartz et al., 2009).

However, logistics are not always in the firm’s control, especially for SMEs. For example, the

quality of roads and transport infrastructure is hardly attributable to the action of the firm,

but rather to the national or even business ecosystem. An impact assessment study of the

Peruvian road network’s expansion between 2003 and 2010 estimates that total Peruvian

exports would have been roughly 20% smaller in 2010 without the road development

programme (Carballo, Volpe Martincus and Cusolito, 2013).

Innovation

Innovative firms have higher levels of productivity and economic growth (Cainelli, Evangelista

and Savona, 2004). They are also more likely to export, and to do it successfully (Love and

Roper, 2013; Cassiman, Golovko and Martínez-Ros, 2010). The capacity to innovate is defined

in different ways: as the ability to generate innovative outputs (Neely et al., 2001) or as the

ability to continuously transform knowledge and ideas into new products, processes and

systems (Lawson and Samson, 2001). In both cases, the capacity to innovate is closely related

to the capability to change.

Innovation, and a firm ability to innovate, is closely related with the technological capacities

of firms. The ability to innovate is particularly important for SMEs (Simon, Houghton and

Aquino, 2000), that are increasingly required to catch up with the rapid advances in new

technologies (Awazu et al., 2009). The wide digitization has also helped SMEs to become more

competitive, as shown by Tanabe and Watanabe (2003) for Japan.

Access to networks, platforms, institutions

In all previous areas of firm’s competency, we have highlighted how forces/determinants

outside the influence of the firm also affect the way firms performs. Management research

highlights the importance of business-to-business networks (Schoonjans, Van Cauwenberge

and Vander Bauwhede, 2013), knowledge sharing, complementarity of resources (Dyer and

Singh, 1998), and effective governance.

Clusters can create links between firms and boost knowledge sharing and positive synergies,

either between firms (business-to-business networks, as for Winters and Stam, 2007) or

between firms and external actors, such as universities or R&D institutes (Acs, Audretsch and

Feldman, 1994). The use of technology in the firm’s network can have positive spillovers on

firms’ performance (Paunov and Rollo, 2016).

Firms also need to be informed about consumers’ needs, demographics and habits, about the

legal requirements they have to comply with, about the status of trade agreements their

country is a signatory of, about the consequences of not being a signatory and the visible and

less visible trade barriers they could encounter if willing to trade. This can be resumed in one

word: connection, the ability to be informed about the nature of and changes in the

competitive environment.

A good connection to the business ecosystem is particularly important for SMEs, which

oftentimes are unable to gather relevant business information (Kitching, Hart and Wilson,

2015; Reid, 1984; Seringhaus, 1987; Christensen, 1991). Help to gather this information

usually comes from public institutions or private associations. But it can also come from

informal institutions, as it is shown in a study conducted in Northern Uganda, where SMEs lack

awareness or the capability to access information from formal trade and investment support

institutions (TISIs) (Okello-Obura et al., 2008).

b. Construction of indices There exist many methodologies to build multidimensional indices, ranging from axiomatic

approaches to multivariate methods. This section reviews some of the most widely used

techniques for index construction.

The most common types of multi-dimensional indices are composite indices. A well-known

example is the Human Development Index (UNDP, 1990-2014) which aggregates through a

geometric mean three dimensions (i.e. life expectancy, education and per capita income)

previously scaled (i.e. by projecting each dimension on a scale from 0 to 1).

Looking at axiomatic approaches, fuzzy sets theory (Zadeh, 1965) has been widely used to

construct indices. The general idea is that membership to a subgroup is determined by a

function allowing for fuzziness (i.e. it may take any value between 0 and 1, rather than 0 or 1

only). Later, the grades of membership in each dimension need to be aggregated, generally

through a weighted arithmetic mean (see for instance Chakravarty, 2006). Several applications

of fuzzy sets theory can be found in the development literature through the measure of

inequality and poverty (see for instance (Basu, 1987); (Chakravarty, 2006); (Shorrocks &

Subramanian, 1994); (Cerioli & Zani, 1990)).

Multivariate methods are another cornerstone to the construction of multidimensional

indices. When modelling multivariate data, researchers tend to think in terms of individual

observations. Taking the regression approach and for instance the least square methodology,

the aim is to minimize the sum of the squared distances between the observed and the

predicted dependent variable for each individual observation. The focus is set on individual

cases, and the relation under study is between the independent variable,𝑦𝑖, and the

dependent variables, 𝑥𝑖.

The Global Competitiveness Index (World Economic Forum 2008-2009) is a good example of

an index that relies on regression methodology. The index incorporates twelve pillars14 of

economic competitiveness. Although the pillars are all meaningful determinants of

competitiveness, their relative importance in explaining competitiveness can vary according

to the specific level of development of each country. To incorporate this fact in the

construction of the final index, the twelve pillars are further regrouped into three sub-pillars

14 The pillars are institutions, infrastructure, macroeconomic stability, health and primary education, higher education and training, good market efficiency, labour market efficiency, financial market sophistication, technological readiness, market size, innovation, business sophistication.

according to different levels of development15: the basic requirements subindex, the efficiency

enhancers subindex and the innovation and sophistication factors subindex. First, specific

weights for each subindex are estimated using maximum likelihood by regressing the level of

GDP per capita on the past values of the subindices. Then, the final index is built from

aggregating through a weighted average the three sub-pillars, for which specific weights have

been estimated according to the stage of development.

Belonging to the literature on latent variables, factor analysis is a well-known statistical

method to handle multivariate data. The aim of factor analysis is to explain a set of observed

variables (i.e. indicators) in terms of a lower number of latent – or unobserved - variables (i.e.

factors). Each observed indicator is treated as a partial manifestation of a postulated broader

latent variable. Uncovering the relationship among the observed indicators allows for the

measurement of the latent concept.

This methodology differs in spirit from classical regression analysis as it emphasizes

covariances rather than individual cases. Additionally, the relationship under study is the one

linking many dependent observed variables with the objective. The aim is to uncover

information about the unobserved independent variable that underlies them. In other words,

the aim is to make use of the relationships among observed indicators to infer something on

the unobserved concept that influences them.

Factor analysis is particularly well suited for the construction of multi-dimensional indices for

various reasons. First, since no indicator is sufficient on its own to predict the underlying latent

variable, factor analysis truly acknowledges multi-dimensionality as essential in the

construction of the final index. Second, factor analysis allows estimating weights (i.e. factor

loadings) associated to each observed indicator in the measurement of the latent factor.

These estimated factor loadings relieve the researcher from subjectively designing the

weighting scheme to follow in the final aggregation step.

Two main types of factor analysis models exist. The first one, Exploratory Factor Analysis (EFA)

does not rely on a particular theoretical model and thus the number of latent variables present

in the data is determined by the pure exploration of the data. Additionally, EFA imposes the

measurement errors to be uncorrelated among them and each indicator to relate to each

latent factor. In contrast, the Confirmatory Factor Analysis (CFA) is based on a pre-specified

theoretical model. CFA allows the research to set in advance the number of latent concepts

as well as which observed indicators are influenced by a specific latent variable. This paper

focuses on CFA, further explained in the next section and in Appendix I.

15 Countries are classified in three stages of development according to two criteria. The first criterion is based on the level of GDP per capita at market exchange rates and the second one is the share of exports of primary goods in total exports to measure the degree to which the economies are factor driven.

3. Confirmatory Factor Analysis

a. Data The confirmatory factor analysis in this paper requires multidimensional data, which cannot

be sourced by a single dataset. Henceforth, this paper uses several datasets.

The standardized World Bank Enterprise Surveys (WBES) is the main source of data for our

paper.16 The WBES dataset reports the answers from enterprise surveys deployed on a

representative sample of formal firms in the non-agricultural sector, by country. Firms are

selected through stratified random sampling (more information on the data can be found in

Dethier, Hirn, & Straub, 2011).

Our analysis retains only the last year available for each country from the cross-section of

firms. We analyse information for 70723 firm observations across 100 countries for the 2006–

14 period. Table 1 reports information on country coverage, while Table 2 summarizes data

coverage across firm size categories, world regions and income levels. It shows that the vast

majority of the countries included in the data we analyse are low and middle income

countries, from all geographic regions. Most firms in the sample are small firms, firms that

report employing less than 20 full-time workers.

The WBES reports the answers to a wide number of questions on firms’ characteristics and

obstacles faced by firms in their activities. We use firm level variables to account for the

capacities of firms to be competitive, and we build proxies for the quality of the business

ecosystem using firm level variables. We build these variables from the WBES, as averages or

shares (depending on the type of variable we use) of firm level answers at the industry j

country c cell, for the latest available year. The choice of the industry-country combination is

motivated by the possibility that, within the same country, different industries are affected

differently by similar issues, and also by the fact that different sectors might perceive the same

issue differently. The industry j is defined using the ISIC code provided in the WBES dataset.

Table 3 in the Appendix provides a description of the variables included in the analysis as well

as their source.

This data is then merged with other macroeconomic datasets from several sources: the World Bank Doing Business Indicators, the World Bank and Turku School of Economics’ Logistics Performance Index, the ISO Survey of Management System Standard Certifications, the World Bank Worldwide Governance Indicators, ITU’s ICT Development Index, UNESCO Institute for Statistics (UIS) and the World Intellectual Property Organization (WIPO). All trade statistics and customs tariff data derive from the ITC Market Analysis Tools.

b. The competitiveness framework Confirmatory factor analysis allows researchers to confirm a model defined a priory. In this

paper, we set up a competitiveness framework based on the review of the literature

conducted in Section 2, which shows how different criteria of competitiveness depend on time

and context (Ambastha and Momaya, 2004). Hence, we organise the different dimensions of

16 Downloaded on January 2016 from http://www.enterprisesurveys.org/data/survey-datasets

http://www.enterprisesurveys.org/data/survey-datasets

competitiveness in the “Competitiveness Grid” (see Figure 1), where we classify the

components of firm competitiveness according to:

How they affect competitiveness: compete, connect and change. These three pillars

reflect traditional static and dynamic notions of competitiveness. The pillars are

reflected in the vertical axis of the grid.

The three layer of the economy at which these components intervene: firm

capabilities, the business ecosystem and the national environment. The layers are

reflected in the horizontal axis of the grid.

How do we populate each cell of the grid in view of the empirical analysis? We draw from the

review of the literature.

a) Compete

i. Firm level: the literature has shown the importance of strong managers, of

meeting quality and sustainability standards and of access to banking services

and inputs for firms to be able to compete and operate today. We proxy these

concepts with the following firm level variables from the WBES: a dummy

indicating if a firm has a quality certification, another dummy for using a bank

account and the years of manager’s experience.

ii. Business ecosystem: the two proxies included in the IBE to enable firms to

compete are the percentage share of firms experiencing power outages and

the percentage share of firms experiencing losses when shipping to domestic

markets, in industry j from country c. These proxies indicate the importance of

a reliable administration of electricity and of a reliable network of suppliers to

be able to operate and timely buy inputs.

iii. National Environment: it provides to the business ecosystem the

macroeconomic framework to operate. We proxy for it with several

macroeconomic indicators from different sources: the ease of getting

electricity (in terms of procedures required), the ease of trading across the

border, the applied tariff rate (to assess how costly it is to import inputs for

production), the logistic performance, the number of quality standards issued

in the country, and the governance index.

b) Connect

i. Firm level: the review of the literature stresses the importance of technology

to be connected with clients and suppliers, and to be aware of the competitors.

At the firm level, we proxy for firm’s capacity to connect with a dummy

indicating if the firm uses email and another dummy for the use of website.

ii. Business ecosystem: we proxy for its quality to support firms’ connectivity with

the share of firms experiencing power outages in industry j in country c. Power

outages, in fact, can hamper the firm’s ability to use ICT.

iii. National environment: we proxy the institutional support provided to

connectivity at the national level with the ITC access score and with the

Government online service score.

c) Change

i. Firm level: access to credit, talent and innovation affect the capacity of firms to

change and remain competitive over time. At the firm level, we proxy for this

with several dummies, indicating if the firm provides training to its employees,

if the firms has financial audit, bank financing and a foreign license.

ii. Business ecosystem: we proxy for its quality with the percentage share of firms

reporting access to finance, business licensing, and an inadequately educated

workforce as an obstacle to their operations.

iii. National environment: to capture how the national framework supports the

business environment, and the firm, we use the ease of getting credit score,

the school life expectancy, the ease of starting a business score, and the

resident patent applications and trademark registrations by country.

c. Empirical framework We specify our econometric model as a Confirmatory Factor Analysis (CFA), as described in (Bollen, 1989) and (Muthén, 1984). 17 The underlying model is presented in Figure 2. In line with our competitiveness framework, we hypothesize a second-order CFA, where the first latent factor is Competitiveness itself measured by three latent sub-concepts: Compete, Connect and Change. We estimate the model following a two-step procedure. First, each pillar (Compete, Connect and Change) is estimated separately, through linear factor analysis. Then, we aggregate the predicted values for each estimated latent pillar (Compete, Connect and Change) into one single measure of competitiveness, through arithmetic mean. As traditional in the factor analysis literature, we estimate the unknown parameters of the model by maximum likelihood. To identify the model, we constrain the factor loading of the first observed indicator to be one. In other words, one unit of change in the latent variable leads to one unit of change in the first observed indicator.

Prediction of the latent scores:

In the case of linear factor analysis, we use the regression method known as the Thompson

method to predict the factor scores. Another method often used in the literature to predict

latent variables in the context of factor analysis is the Bartlett’s factor score.

There has been a long debate in the literature on which prediction method is best. Since each

method has some desirable properties, there is no clear answer. For instance, the Bartlett’s

factor score is an unbiased estimate of the latent variable, but it suffers from being less

accurate in terms of average prediction error, compared to the Thompson’s score.

When we apply the nonlinear factor analysis, we use the empirical Bayes method to predict

latent factor scores.

d. Results We report the results from the estimation of the factor analysis specified as for Figure 2. We

estimate each pillar (Compete, Connect and Change) separately, through linear factor

17 For more details see Appendix.

analysis. We then predict values for Compete, Connect and Change and aggregate them into

one index of competitiveness through an arithmetic mean.

To deal with the substantial amount of missing values, we propose to use a full information

maximum likelihood method implemented in Stata 14 (StataCorp, 2015) as an option to the

sem command. This technique assumes joint normality of all variables as well as the missing

values to be missing at random (MAR) so that maximum likelihood can be coupled with a

simple imputation procedure.

The estimation results of the Competitiveness path diagram are displayed in Table 4. All the

coefficients are reported in their standardized forms with their corresponding robust standard

errors in parenthesis.

Focusing on our first latent concept, Compete, we see that all the estimated coefficients (i.e.

the factor loadings) are of expected sign and significant at the 1% level. Notably, all the

variables are positively associated with the Compete pillar except for: the share of firms

experiencing power outages (Power Outages), the share of firms affected by losses when

shipping to domestic markets (Shipping losses) or the rate of tariff on imports (Applied tariff

rate). This is an indication that the results are in line with expectations, because increase in

the indicators that are negatively associated with Compete (like Power Outages) means that

more firms complain about experiencing problems with the business ecosystem, like having

power outages, an element which usually cuts or reduces production and daily activities at

the level of any enterprise. Since all indicators related with the business ecosystem identify

obstacles or constraints, these indicators should not positively be associated with any of the

pillars of competitiveness. The coefficient of Applied tariffs is also negative as expected: higher

tariffs on imported goods are an obstacle to the purchase of inputs.

With regard to the second latent concept, Connect, the variables measuring an enhanced

connectivity – for instance whether a firm uses emails or a website to communicate with

suppliers or clients - are positively associated with the latent variable, whereas the share of

firms reporting to have experienced electricity as an obstacle to their operations is negatively

correlated with our Connect pillar. Once again this indicates that the framework proposed is

working in line with expectations and economic literature and intuition.

Finally, the last column of Table 4 summarizes the estimation results associated with the third

pillar, Change. Again, we see that all the coefficients are of expected sign and significant at the

1% level.

As a robustness check, we also estimate the whole model at once, instead of estimating it using a two steps procedure. The coefficients, in line with previous results, are reported in Table 5. Finally, to account for the fact that the model includes both continuous and binary variables, we also perform a nonlinear factor analysis, as described in (Muthén, 1984). The results, qualitatively similar to those from the linear factor analysis, are reported in Table 6.

Based on the sign of the coefficients as well as their significance in Tables 4 to 6, we can

conclude that the variables chosen in each pillars are measuring our concept of Compete,

Connect and Change.

4. Relevance of the Competitiveness Index

In order to verify that our indices for Compete, Connect and Change, as well as our final index

of Competitiveness, are good measures, we regress each index on a battery of firm i proxies

of competitiveness (𝑧𝑖), those mainly used in the literature: labour productivity (windsorized,

so as to reduce the outlier bias), the percentage of inputs of foreign origin used by the firm,

the share of total sales that are exported, and the exporting status.

Table 7 presents the estimation results from the regression of the predicted values for

Compete (𝐶𝑖1), Connect (𝐶𝑖

2) and Change (𝐶𝑖3) (obtained through CFA as described in Section

3), on the proxies of competitiveness.

Equation 1 𝒛𝒊 = 𝜶 + 𝜷𝟏 ∗ 𝑪𝒊𝟏 + 𝜷𝟐 ∗ 𝑪𝒊

𝟐 + 𝜷𝟑 ∗ 𝑪𝒊𝟑 + 𝜸𝒋 + 𝜸𝒄 + 𝜺𝒊

The regression (as per Equation 1) includes country (𝛾𝑐) and sector (𝛾𝑗) fixed effects, to control

for country c and sector j characteristics that affect all firms within the same country or sector

equally, and has robust standard errors. We find a positive and significant correlation between

the three predicted values for Compete, Connect and Change and the main proxies of

competitiveness (𝑧𝑖).

We then regress the Competitiveness index (𝐶𝐼𝑖) (built as the arithmetic mean of the three

pillars) on the main proxies of competitiveness.

Equation 2 𝒛𝒊 = 𝜶 + 𝜹𝟏 ∗ 𝑪𝑰𝒊 + 𝜸𝒋 + 𝜸𝒄 + 𝜺𝒊

Equation 3 𝒛𝒊 = 𝜶 + 𝜹𝒆𝒙𝒑 ∗ 𝑪𝑰𝒊 ∗ 𝒆𝒙𝒑 + 𝜹𝒏𝒆𝒙𝒑 ∗ 𝑪𝑰𝒊 ∗ 𝒏𝒆𝒙𝒑 + 𝜸𝒋 + 𝜸𝒄 + 𝜺𝒊

Equation 4 𝒛𝒊 = 𝜶 + 𝜹𝑺 ∗ 𝑪𝑰𝒊 ∗ 𝑺 + 𝜹𝑴 ∗ 𝑪𝑰𝒊 ∗ 𝑴 + 𝜹𝑳 ∗ 𝑪𝑰𝒊 ∗ 𝑳 + 𝜸𝒋 + 𝜸𝒄 + 𝜺𝒊

Once again, we include country and sector fixed effects, and standard errors are robust (as

per Equation 2). Table 8 shows that the index is positively and significantly correlated with all

proxies. Interestingly, when in column (7) we differentiate between exporting and non-

exporting firms (as per Equation 3), results are maintained for both types of firms. Similarly,

when we split firms by size (as per Equation 4) in columns (8-10), results apply to firms of all

sizes. These results provide further evidence both of the fact that our index is a valid measure

of competitiveness, and that our proposed framework of competitiveness applies to all firms,

independently of their exporting status and of their size.

Finally, we try to verify the quality of our indices by conducting some graphic analysis. Since it

is reasonable to assume that firms in low income countries will be less competitive than firms

in high income countries, on average, we plot the predicted values for Compete, Connect and

Change, as well as the Competitiveness Index (normalized between 0-100 and averaged by

country) on GDP per capita, as for Figure 3. The plots confirm that firms in richer countries

perform better, as expected.

Most importantly, Figure 4 shows that the performance gap between large and small firms is

higher in lower income countries than in richer countries. This finding is supported by several

reports18, and notably by data available for Latin American and European countries that have

been reported by McDermott, Gerald A. and Pietrobelli, Carlo (2015) in an ITC working

paper19.

The positive relationship between our index of Competitiveness and labour productivity, the

classic proxy for competitiveness is confirmed in the plot in Figure 5.

5. Conclusive remarks

Competitiveness is a multidimensional concept, not easy to define or calculate. Summarizing

several dimensions of competitiveness into one single measure is a challenging task, but

important and worth trying, since it can allow policy makers to monitor not only the health of

their firms but also the efficiency of the policies put in place to help them.

Competitiveness is not a new concept, but to date productivity remains the most commonly

used way to measure it, at both the macro and micro level. However, whether productivity

fully represents the performance or competitive strength of firms or countries remains a

subject of discussion.

This paper proposes to measure competitiveness by shaping its multi-dimensionality into an

index of firm competitiveness. The first contribution of this paper is therefore of filling a gap

in the attempt to measure competitiveness, until now mainly proxied with several and open

to discussion measures of productivity. It does so by proposing to measure competitiveness

using confirmatory factor analysis.

In order to summarize multidimensional realities into one single measure of competitiveness,

we conceptualise a framework to capture this multi-dimensionality. Therefore, the second

contribution of this paper stays in proposing and testing a competitiveness framework, based

on the review of the economic and management literature.

Our results suggest that the Competitiveness Index from our confirmatory factor analysis is

positively correlated with commonly used proxies of competitiveness, such as labour

18 a) SME Competitiveness Outlook: Connect, Compete and Change for Inclusive Growth (2015). International Trade Centre, Geneva b) Perspectives on Global Development: Boosting Productivity to meet the middle-income challenge (2014). OECD, Paris. c) On the role of productivity and factor accumulation in economic development in Latin America and the Caribbean (2010). Inter-American Development Bank. 19 McDermott, Gerald A. and Pietrobelli, Carlo (2015). SMEs, Trade and Development in Latin America: Toward a new approach on Global Value Chain Integration and Capabilities Upgrading. ITC Working paper. International Trade Centre, Geneva

productivity, the probability to export, the percentage of inputs of foreign origin used by the

firm and the share of total sales that were exported.

The multidimensional framework we build proves to be applicable to firms of different size

and to both exporting and non-exporting firms, as shown by the positive relationship between

labour productivity and the index for the different types of firms. As expected, firms in richer

countries perform better than firms in low income countries, independently of firm’s size.

Interestingly, the performance gap between small and large firms is higher in lower income

countries than in richer countries.

Even though further research on measuring competitiveness is needed, our paper proposes

an alternative framework of competitiveness and a way to test for it. It is the starting point

for further research, on both the empirical and the theoretical side. In fact, future research

could focus on assessing whether the framework proposed in our paper can be tested using

different statistical techniques. Finally, theoretical work in the area of firm competitiveness

should be developed to combine the dynamic and static nature of competitiveness, as well as

to integrate the business environment of the firm into the complex and multidimensional

system of forces that shape firms’ performance, position and direction.

References

Basu, K. (1987). Axioms for fuzzy measure of inequality. (Elsevier, Ed.) Mathematical Social

Sciences, 14(3), 275--288.

Bollen, A. K. (1989). Structural equations with latent variables. (W. Interscience, Ed.) Wiley

series in probability and mathematical statistics.

Cerioli, A., & Zani, S. (1990). A fuzzy approach to the measurement of poverty. (Springer, Ed.)

Income and wealth distribution, inequality and poverty, 272--284.

Chakravarty, S. R. (2006). An axiomatic approach to multidimensional poverty measurement

via fuzzy sets. A. Lemmi and G. Betti (eds.) Fuzzy Set Approach to Multidimensional

Poverty Measurement, Springer-Vrlag, New York.

Muthén, B. (1984). A general structural equation model with dichotomous, ordered

categorical, and continuous latent variable indicators. (Springer, Ed.) Psychometrika,

49(1), 115--132.

Shorrocks, A. F., & Subramanian, S. (1994). Fuzzy poverty indices. University of Essex.

StataCorp. (2015). Stata Structural Equation Modeling. (C. S. Press, Ed.) Reference Manual

Release 13.

UNDP. (1990-2014). Statistics - Huamn Development Reports. Human Development Reports.

Retrieved from http://hdr.undp.org/en/statistics/

Zadeh, L. A. (1965). Fuzzy sets. Information and control, 338--353.

ITC (2015). SME Competitiveness Outlook: Connect, compete and change for inclusive growth.

International Trade Centre. Geneva.

Garelli, Stephane (2006). Top Class Competitors. How nations, firms, and individuals succeed in the new

world of competitiveness. John Wiley & Sons, Ltd.

McMillan, John (2008). Market institutions. In The New Palgrave Dictionary of Economics, 2nd edn,

Larry Blume and Steven Durlauf, ed. Palgrave Macmillan UK.

Biggs, Tayler ( ??). Is Small Beautiful and Worthy of Subsidy? Leterature review. Mimeo. Available at

http://www.enterprise-development.org/wp-

content/uploads/Is_Small_Beautiful_and_Worthy_of_Subsidy.pdf

Fafchamps, Marcel (2004). Market Institutions in Sub-Saharan Africa. Theory and Evidence. Cambridge:

MIT Press.

Khanna, Tarun, Krishna G. Palepu and Jayant Sinha (2005). Strategies that fit emerging markets.

Harvard Business Review, Issue 83 (June), pp. 4–19.

World Economic Forum (2008). The Global Competitiveness Report 2008-2009. Geneva. Available at

https://www.weforum.org/reports/global-competitiveness-report-2008-2009/

Álvarez, Isabel, Raquel Marín and Georgina Maldonado (2009). Internal and external factors of

competitiveness in the middle-income countries. Working Papers Series; No. 08/09. Complutense

University of Madrid. Available from: http://eprints.ucm.es/9570/

Porter, Michael E. The Five Competitive Forces That Shape Strategy. Harvard Business Review, vol. 86, No. 1 (January), pp. 78–93. Barney, Jay (1991). Firm Resources and Sustained Competitive Advantage. Journal of Management, vol. 17, No. 1, pp. 99-120. Dunning, John H. and Sarianna M. Lundan (2010). The institutional origins of dynamic capabilities in multinational enterprises. Industrial and Corporate Change, vol. 19, No. 4, pp. 1225–1246. Augier, Mie and David J. Teece (2007). Dynamic capabilities and multinational enterprise: Penrosean insights and omissions. Management International Review, 47(2), pp. 175–192.

Paunov, Caroline and Valentina Rollo (2016). Has the Internet Fostered Inclusive Innovation in the

Developing World? World Development, vol. 78, pp. 587-609.

Dyer, Jefrey H. and Harbir Singh (1998). The Relational View: Cooperative Strategy and Sources of

Interorganizational Competitive Advantage. Academy of Management Review, vol. 23, No. 4, pp. 660-

679.

Schoonjans, Bilitis, Philippe Van Cauwenberge and Heidi Vander Bauwhede (2013). Formal Business

Networking and SME growth. Small Business Economics, vol. 41, Issue 1 (June), pp.169-181.

Harinder, Singh, Jaideep Motwani and Ashok Kumar (2000). A review and analysis of the state-of-the-art research on productivity measurement. Industrial Management & Data Systems, vol. 100, Issue 5, pp. 234 – 241. Tangen, Stefan (2005), Demystifying productivity and performance. International Journal of Productivity and Performance Management, vol. 54, Issue 1, pp. 34 – 46. OECD (2001). Measuring Productivity. Measurement of Aggregated and Industry-Level Productivity Growth. OECD Manual, OECD Publishing, Paris. Available at https://www.oecd.org/std/productivity-stats/2352458.pdf. OECD (2016). OECD Compendium of Productivity Indicators 2016. OECD Publishing, Paris. Available at http://www.oecd.org/std/productivity-stats/oecd-compendium-of-productivity-indicators-22252126.htm. Bris, Arturo and José Caballero (2016). Revisiting the Fundamentals of Competitiveness: A proposal. In IMD World Competitiveness Yearbook 2016. IMD World Competitiveness Center, Lausanne. Feurer, Rainer and Kazem Chaharbaghi (1994). Defining Competitiveness: A holistic approach. Management Decision, vol. 32, No. 2, pp. 49-58. Pakes, Ariel (2015). Empirical Tools and Competition Analysis: Past Progress and Current Problems. Mimeo. Available at http://scholar.harvard.edu/files/pakes/files/empiricaltools-10-2015.pdf?m=1444080861.

Nelson, Richard, R. (1996). The source of Economic Growth. Harvard University Press. Cambridge, Massachusttes/London, England. Bertrand, Marianne and Antoinette Schoar (2003). Managing with Style: The effect of managers on

firm policies. The Quarterly Journal of Economics, vol. 118, Issue 4 (November), pp. 1169-1208.

Brunninge, Olof, Mattias Nordqvist and Johan Wiklund (2007). Corporate Governance and Strategic

Change in SMEs: The Effects of Ownership, Board Composition and Top Management Teams. Small

Business Economics, vol. 29, Issue 3 (October), pp. 295-308.

Carpenter, Mason A. and James W. Fredrickson (2001). Top Management Teams, Global Strategic

Posture, and the Moderate Role of Uncertainty. Academy of Management Journal, vol. 44, No. 3, pp.

533-545.

Hambrick, Donald C. and Phyllis A. Mason (1984). Upper Echelons: The Organization as a Reflection of

Its Top Managers. Academy of Management Review, vol. 9, No. 2, pp. 193-206.

Hambrick, Donald C (2007). Upper Echelons Theory: An update. Academy of Management Review, vol.

32, No. 2, pp. 334-343.

Goedhuysa, Micheline and Leo Sleuwaegen (2016). International standards certification, institutional

voids and exports from developing country firms. International Business Review, in press, available

online 6 May 2016.

http://www.oecd.org/std/productivity-stats/oecd-compendium-of-productivity-indicators-22252126.htmhttp://www.oecd.org/std/productivity-stats/oecd-compendium-of-productivity-indicators-22252126.htm

Hudson, John and Philip Jones (2003). International trade in “Quality Goods”? Signalling Problems for

Developing Countries. Journal of International development, vol. 15, Issue 8, pp. 999–1013.

Berman, Nicolas and Jérôme Héricourt (2010). Financial factors and the margins of trade: Evidence from cross-country firm-level data. Journal of Development Economics, vol. 93, Issue 2 (Novmber), pp. 206-217.

Medina, Leonardo (2012). Spring Forward or Fall Back? The Post-Crisis Recovery of Firms. IMF Working Paper, No. 12/292, International Monetary Fund. Available at https://www.imf.org/external/pubs/ft/wp/2012/wp12292.pdf

Braun, Matias and Borja Larrain (2005). Finance and the Business Cycle: International, Inter-Industry Evidence. The Journal of finance, vol. 60, No. 3, pp. 1097-1128.

Levine, Ross and Zervos, Sara. "Stock Markets and Economic Growth." American Eco- nomic Review, June 1998, 88(3), pp. 537- 58.

Rajan, Raghuram G. and Luigi Zingales (1998). Financial development and Growth. American Economic Review, vol. 88, No. 3 (June), pp. 559-586.

Coricelli, Fabrizio, Nigel Driffield, Sarmistha Pal and Isabelle Roland (2010). Excess Leverage and Productivity Growth in Emerging Economies : Is There a Threshold Effect ? Discussion Paper, No. 4834 (March). The Institue for the Study of Labour (IZA), Bonn, Germany. Available at http://ftp.iza.org/dp4834.pdf.

Backman, Mikaela (2014). Human capital in firms and regions: Impact on firm productivity. Papers in Regional Science, vol. 93, Issue 3 (August), pp. 557–575.

OECD (2013), Skills Development and Training in SMEs, Local Economic and Employment Development

(LEED), OECD Publishing. Available at

http://www.skillsforemployment.org/wcmstest4/groups/skills/documents/skpcontent/ddrf/mdu0/~

edisp/wcmstest4_054646.pdf

Amiti, Mary and Josef Konings (2007), Trade Liberalization, Intermediate Inputs, and Productivity:

Evidence from Indonesia. The American Economic Review, vol. 97, No. 5 (December), pp. 1611-1638.

Şeker, Murat (2012). Importing, Exporting, and Innovation in Developing Countries. Review of

International Economics, vol. 20, Issue 2 (May), pp. 299-314.

Atalay, Murat, Nilgün Anafarta and Fulya Sarvan (2013). The relationship between innovation and firm

performance: An empirical evidence from Turkish automotive supplier industry. Procedia - Social and

Behavioral Science, vol. 75, pp. 226-235.

Cassiman, Bruno, Golovko, Elena and Ester Martínez-Ros (2010). Innovation, exports and productivity.

International Journal of Industrial Organization, vol. 28, Issue 4 (July), pp. 372-376.

Rubera, G. and Kirca, A., (2012), Firm innovativeness and its performance outcomes: A meta-analytic

review and theoretical integration, Journal of Marketing, 76(3), pp.130-147.

http://onlinelibrary.wiley.com/doi/10.1111/pirs.2014.93.issue-3/issuetoc

Appendix I: Tables and Figures

Figures Figure 1: The Competitiveness Grid

Competitiveness Grid

Pillars

Capacity to compete Capacity to connect Capacity to change

Laye

rs

‘Firm level’ capabilities

Business ecosystem

National environment

Figure 2: Competitiveness Path Diagram where observed variables are indicated by rectangles, latent variables by ellipses and measurement errors by circles.

Figure 3: Competitiveness Indices by income

020

40

60

80

10

0

Co

mpe

titiven

ess

0 10000 20000 30000GDP per capita (PPP$)

avg_scoreM Fitted values

Competitiveness vs GDP

02

04

06

08

01

00

Com

pete


avg_scorecomp Fitted values

Compete vs GDP

02

04

06

08

01

00

Con

ne

ct


avg_scoreconn Fitted values

Connect vs GDP

02

04

06

08

01

00

Cha

ng

e


avg_scorech Fitted values

Change vs GDP

Figure 4: Competitiveness Indices by income: Gap between Large and Small firms

Figure 5: Competitiveness Indices versus Labour Productivity

.51

1.5

22.5

3

Co

mpe

titiven

ess G

ap (

Larg

e m

inus S

mall)


m_gapLS Fitted values

Competitiveness Gap vs GDP

020

40

60

80

10

0

Co

mpe

titiven

ess

0 1 2 3 4 5Labour Productivity

avg_scoreM Fitted values

Competitiveness vs Labour Productivity

Tables Table 1: Data coverage by country and year

Country Year Observations Percentage share

in tota l


in tota l


in tota l

Angola 2010 360 0.509 Indones ia 2009 1444 2.042 Poland 2013 542 0.766

Albania 2013 360 0.509 India 2014 9281 13.123 Paraguay 2010 361 0.51

Argentina 2010 1054 1.49 Israel 2013 483 0.683 Romania 2013 540 0.764

Armenia 2013 360 0.509 Jamaica 2010 376 0.532 Russ ian Federation 2012 4220 5.967

Azerbai jan 2013 390 0.551 Jordan 2013 573 0.81 Rwanda 2011 241 0.341

Burundi 2014 157 0.222 Kazakhstan 2013 600 0.848 Senegal 2014 601 0.85

Burkina Faso 2009 394 0.557 Kenya 2013 781 1.104 Sierra Leone 2009 150 0.212

Bangladesh 2013 1442 2.039 Kyrgyz Republ ic 2013 270 0.382 El Sa lvador 2010 360 0.509

Bulgaria 2013 293 0.414 Cambodia 2013 472 0.667 Serbia 2013 360 0.509

Bol ivia 2010 362 0.512 Lao PDR 2012 270 0.382 Suriname 2010 152 0.215

Brazi l 2009 1802 2.548 Lebanon 2013 561 0.793 Slovak Republ ic 2013 268 0.379

Barbados 2010 150 0.212 Sri Lanka 2011 610 0.863 Slovenia 2013 270 0.382

Botswana 2010 268 0.379 Lesotho 2009 151 0.214 Sweden 2014 600 0.848

Chi le 2010 1033 1.461 Li thuania 2013 270 0.382 Swazi land 2006 307 0.434

China 2012 2700 3.818 Latvia 2013 336 0.475 Chad 2009 150 0.212

Cote d'Ivoire 2009 526 0.744 Morocco 2013 407 0.575 Tajikis tan 2013 359 0.508

Cameroon 2009 363 0.513 Moldova 2013 360 0.509 Timor-Leste 2009 150 0.212

Colombia 2010 942 1.332 Madagascar 2013 532 0.752 Trinidad and Tobago 2010 370 0.523

Cape Verde 2009 156 0.221 Mexico 2010 1480 2.093 Tunis ia 2013 592 0.837

Costa Rica 2010 538 0.761 Macedonia 2013 360 0.509 Turkey 2013 1344 1.9

Czech Republ ic 2013 254 0.359 Mal i 2010 360 0.509 Tanzania 2013 813 1.15

Dominican Republ ic 2010 360 0.509 Myanmar 2014 632 0.894 Uganda 2013 762 1.077

Egypt 2013 2897 4.096 Montenegro 2013 150 0.212 Ukra ine 2013 1002 1.417

Estonia 2013 273 0.386 Mongol ia 2013 360 0.509 Uruguay 2010 607 0.858

Ethiopia 2011 644 0.911 Mozambique 2007 479 0.677 Venezuela 2010 320 0.452

Gabon 2009 179 0.253 Mauri tania 2014 150 0.212 Vietnam 2009 1053 1.489

Georgia 2013 360 0.509 Mauri tius 2009 398 0.563 Yemen 2013 353 0.499

Ghana 2013 720 1.018 Malawi 2014 523 0.74 South Africa 2007 937 1.325

Guinea 2006 223 0.315 Nigeria 2014 2676 3.784 Zambia 2013 720 1.018

Gambia 2006 174 0.246 Nicaragua 2010 336 0.475 Zimbabwe 2011 599 0.847

Guatemala 2010 590 0.834 Nepal 2013 482 0.682

Guyana 2010 165 0.233 Pakis tan 2013 1247 1.763

Honduras 2010 360 0.509 Panama 2010 365 0.516

Croatia 2013 360 0.509 Peru 2010 1000 1.414

Hungary 2013 310 0.438 Phi l ippines 2009 1326 1.875

Table 2: Data coverage by firm size, sector, income level and world region

Group Observations Percentage share in total

Size Category

small (

Table 3: Description of variables used in the confirmatory factor analysis

Variable name Mean sd Source

Firm-level capabilities

Qual i ty certi fication 0.26

Bank account 0.87

Manager's experience 2.68

[17]

0.67

Emai l 0.74

Webs ite 0.50

Tra ining 0.40

Fiancia l audit 0.55

Bank financing 0.35

Foreign l icences 0.14

Power outages 59.16 22.16

Shipping losses 17.19 11.86

Obstacle: electrici ty 47.40 21.18

Access to finance constra int 45.22 17.81

Licens ing constra int 30.55 16.44

Inadequate workforce

education

39.14 20.48

Getting electrici ty 63.72

Trading across boarders 58.00

Appl ied tari ff rate 0.09 0.04 ITC, based on data from ITC Market Analys is

Tools , 2006–2015

(www.intracen.org/marketanalys is ).

Logis tic performance 2.89 World Bank and Turku School of Economics ,

Logis tics Performance Index 2014,

http://lpi .worldbank.org/

Immediate buisness environment

National environment

Appl ied tari ff rate, trade-weighted mean, a l l products (%).A tari ff i s a customs duty that i s levied by

the destination country on imports of merchandise goods .

Trade-weighted average tari ff i s ca lculated for each importing country us ing the trade patterns of

the importing country’s reference group (based on 2013 trade s tatis tics ). To the extent poss ible,

speci fic rates have been converted to their ad va lorem equiva lent rates and included in the

ca lculation of weighted mean tari ffs . Preferentia l tari ff arrangements (tari ff preferences) have been

taken into account.

A multidimens ional assessment of logis tics performance, the Logis tics Performance Index (LPI),

compares the trade logis tics profi les of 160 countries and rates them on a sca le of 1 (worst) to 5

(best). The ratings are based on 6,000 individual country assessments by nearly 1,000 international

freight forwarders , who rated the eight foreign countries their company serves most frequently.

Percentage share of fi rms identi fying an inadequately educates workforce as an obstacle to their

current operations .

Doing Bus iness ‘Ease of getting electrici ty’ score (0–100). Al l procedures required for a bus iness to

obtain a permanent electrici ty connection and supply for a s tandardized warehouse.

World Bank, International Finance Corporation,

Doing Bus iness 2014:

Understanding Regulations for Smal l and

Medium-Size Enterprises ,

http://www.doingbus iness .org/

methodologysurveys/Doing Bus iness ‘Ease of trading across borders ’ score (0–100). The indcator measures the time and

cost (excluding tari ffs ) associated with exporting and

importing a s tandardized cargo of goods by sea transport.

A dummy equals to one i f the fi rm has a l ine of credit or a loan from a financia l insti tution.

A dummy equals to one i f the fi rm uses technology l icensed from a foreign-owned company,

Percentage share of fi rms experiencing power outages in industry j of country c. Authors ' own ca lculation;

Fi rm level data source:

Enterprise Surveys

(http://www.enterprisesurveys .org),

The World Bank (2005–2014)

Percentage share of fi rms experiencing losses when shipping to domestic markets in industry j of

Percentage share of fi rms experiencing electrici ty as being an obstacle to their current operations .

Percentage share of fi rms reporting access to finance as an obstacle to their current operations .

Percentage share of fi rms identi fying buisness l icens ing and permits as an obstacle to their current

Description

A dummy equals to one i f the fi rm has an international ly-recognized qual i ty certi fication. Enterprise Surveys

(http://www.enterprisesurveys .org),

The World Bank (2005–2014)

A dummy equals to one i f the fi rm has a checking or savings account.

Logari thm of years of the managers ’ experience [years of managers experience]

A dummy equals to one i f the fi rm uses emai l to communicate with cl ients or suppl iers

A dummy equals to one i f the fi rm has i ts own webs ite.

A dummy equals to one i f the fi rm offers formal tra ining programs for i ts permanent, ful l -time

A dummy equals to one i f the fi rm had i ts annual financia l s tatements checked and certi fied by an

external auditor.

ISO qual i ty s tandards 21,386 65,497 ISO, The ISO Survey of Management System

Standard Certi fications , 2013, www.iso.org

Governance -0.37 World Bank, Worldwide Governance Indicators

(2014),

http://info.worldbank.org/governance/wgi/inde

x.aspx#home

ICT Access 4.73 ITU, Measuring the Information Society 2014, ICT

Development Index 2014 (2013 data except for

Ta jikis tan, 2008),

http://www.itu.int/en/ITU-

D/Statis tics/Pages/publ ications/mis2014.aspx

Government onl ine service 0.47 UNPAN, e-Government Survey 2014, http://

www2.unpan.org/egovkb/

Getting credit 19.87 World Bank, Ease of Doing Bus iness Index 2014,

Doing Bus iness 2014,

http://www.doingbus iness .org/reports/global -

reports/doing-bus iness-2014

School l i fe expectancy 12.56 2.46 UNESCO Insti tute for Statis tics (UIS), 2001–2013,

http://stats .uis .unesco.org

Starting a bus iness 80.36 World Bank, Ease of Doing Bus iness Index 2014,

Doing Bus iness 2014,

http://www.doingbus iness .org/methodology/st

arting-a-bus iness

Patent appl ications 65.47 141.44 WIPO, 2000–2013,

http://www.wipo.int/porta l/en/index.html

Trademark regulations 611.93 654.1 WIPO, 2004–2013,

http://www.wipo.int/porta l/en/index.html

Doing Bus iness ‘Ease of getting credit’ score (0–100). The index measures the legal rights of

borrowers and lenders with respect to secured transactions through one set of indicators and the

sharing of credit information through another.

School l i fe expectancy, primary to tertiary education (years ). Total number of years of school ing that

a chi ld of a certa in age can expect to receive in the future, assuming that the probabi l i ty of his or

her being enrol led in school at any particular age is equal to the current enrolment ratio for that

age.

Doing Bus iness ‘Ease of s tarting a bus iness ’ score (0–100). The index measures the number of

procedures , time and cost for a smal l and medium-s ize l imited l iabi l i ty company to s tart up and

formal ly operate.

Res ident patent appl ications , equiva lent count by appl icant’s origin (per mi l l ion people). Patent

fi l ings made by appl icants at their home office (national or regional ), a lso ca l led domestic

appl ications . Appl ications at regional offices are equiva lent to multiple appl ications , one in each

of the s tate members of those offices , therefore each appl ication is multipl ied by the

corresponding number of member s tates , except for the European patent Office (EPO) and the

African Regional Intel lectual Property Organization (ARIPO), for which des ignated countries are not

known, in which case each appl ication is counted as one appl ication abroad i f the appl icant does

not res ide in a member s tate; or as one res ident and one appl ication abroad i f the appl icant

res ides in a member s tate.

Res ident trademark regis trations , equiva lent class count by appl icant’s origin (per mi l l ion people).

Number of "ISO 9001:2008 Qual i ty management systems" certi ficates i ssued (per mi l l ion people).

Governance index. Average score over s ix dimens ions of governance: voice and accountabi l i ty,

pol i tica l s tabi l i ty and absence of violence, government effectiveness , regulatory qual i ty, rule of law,

and control of corruption.

ICT access sub-index score (0–10). Compos ite index that weights five ICT indicators (20% each): (1)

Fixed-telephone subscriptions per 100 inhabitants ; (2) Mobi le-cel lular telephone subscriptions per

100 inhabitants ; (3) International Internet bandwidth (bi t/s ) per Internet user; (4) Percentage of

households with a computer; and (5) Percentage of households with Internet access .

Government’s onl ine service index score (0-1). Each country’s national webs ite i s assessed for

content, features , access ibi l i ty and uptake, including the national centra l porta l , e-services porta l ,

and e-participation porta l as wel l as the webs ites of the related minis tries of education, labour,

socia l services , health, finance, and environment, as appl icable.

29

Table 4: Estimation results for the linear factor analysis by pillar.

Components of Competitveness by Pillar

Compete Connect Change

Firm

leve

l

Quality certification 0.130*** (0.0042 )

Email 0.426*** (0.0044)

Training 0.169*** (0.0043)

Bank account 0.166*** (0.0045)

Website 0.369*** (0.0045)

Financial audit 0.022*** (0.0044)

Manager's experience 0.149*** (0.0041)

Bank financing 0.154*** (0.0042)

Foreign licences 0.030*** (0.0054)

Bu

sin

ess

eco

syst

em

Power outages -0.636*** (0.0031)

Obstacle: electricity

-0.593*** (0.0029)

Access to finance constraint

-0.501*** (0.0047)

Shipping losses -0.111*** (0.0060)

Licensing constraint

-0.458*** (0.0053)

Inadequate workforce education

-0.049*** (0.0066)

Nat

ion

al E

nvi

ron

me

nt

Getting electricity 0.721*** (0.0029)

ICT Access 0.802*** (0.0026)

Getting credit 0.421*** (0.0039)

Trading across boarders

0.745*** (0.0033)

Government online service

0.712*** (0.0027)

School life expectancy

0.838*** (0.0021)

Applied tariff rate -0.562*** (0.0030)

Starting a business

0.447*** (0.0039)

Logistic performance 0.562*** (0.0032)

Patent applications

0.604*** (0.0035)

ISO quality standards 0.093*** (0.0020)

Trademark regulations

0.839*** (0.0030)

Governance 0.842*** (0.0022)

Observations 70723 70723 70723

Robust standard errors in parentheses

*** p

30

Table 5 : Estimation results linear factor analysis on the whole model

Components of Competitiveness


Firm

leve

l

Quality certification 0.130*** (0.0041 )

Email 0.426*** (0.0038)

Training 0.169*** (0.0041)

Bank account 0.166*** (0.0042)

Website 0.369*** (0.0039)

Financial audit 0.022*** (0.0042)


Bank financing 0.154*** (0.0041)

Foreign licences 0.029*** (0.0049)

Bu

sin

ess

eco

syst

em

Power outages -0.636*** (0.0030)


-0.593*** (0.0032)


-0.501*** (0.0035)



-0.458*** (0.0038)


-0.049*** (0.0045)

Nat

ion

al E

nvi

ron

me

nt


ICT Access 0.802*** (0.0026)

Getting credit 0.421*** (0.0039)


0.745*** (0.0026)


0.712*** (0.0027)


0.838*** (0.0019)


Starting a business

0.447*** (0.0038)


Patent applications

0.604*** (0.0035)



0.839*** (0.0021)

Governance 0.842*** (0.0018)



*** p

31

Table 6: Estimation results of non-linear factor analysis by pillar

Components of Competitveness by Pillar


Firm

leve

l

Quality certification 1 (constrained) Email 1 (constrained)

Training 1 (constrained)

Bank account 1.831*** (0.0741)

Website 0.369*** (0.0045)

Financial audit

0.128*** (0.0244)


Bank financing

0.931*** (0.0335)

Foreign licences

0.237*** (0.0429)

Bu

sin

ess

eco

syst

em

Power outages -49.31*** (1.5866)


-10.92*** (0.1719)


-25.62*** (0.7346)



-21.60*** (0.6609)


-2.912*** (0.4037)

Nat

ion

al E

nvi

ron

me

nt


ICT Access 1.197*** (0.0187)

Getting credit

24.32*** (0.7090)


53.86*** (1.7674)


0.118*** (0.0017)


5.851*** (0.1604)


Starting a business

13.10*** (0.4006)


Patent applications

254.2*** (6.6163)



1622.3*** (43.378)

Governance 1.611*** (0.0518)



*** p

32

Table 7 : Regression results by pillar, with country and sector fixed effects

(1) (2) (3) (4) (5) (6)

VARIABLES

ln(Lab Prod usd) wind

Percentage of imported

inputs

Percentage of sales

exported

Exporter Exporter Exporter

LPM Logit Margin

Compete 0.041*** 1.112*** 0.681*** 0.021*** 0.127*** 0.019***

(0.007) (0.206) (0.128) (0.002) (0.014) (0.002)

Connect 0.062*** 1.107*** 1.028*** 0.023*** 0.183*** 0.027***

(0.003) (0.069) (0.042) (0.001) (0.006) (0.001)

Change 0.087*** 1.439*** 1.096*** 0.032*** 0.215*** 0.032***

(0.006) (0.159) (0.100) (0.002) (0.011) (0.002)

Observations 23,351 16,248 26,453 26,546 26,546 26,546

R-squared 0.226 0.254 0.126 0.175


*** p

33

Table 8 : Regression results for the competitiveness index (arithmetic mean) country and sector fixed effects.

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

VARIABLES

ln(Lab Prod usd)

wind

Percentage of imported

inputs

Percentage of sales

exported

Exporter Exporter Exporter ln(Lab Prod usd)

wind

ln(Lab Prod usd)

wind

Percentage of

imported inputs

Percentage of sales

exported

LPM Logit Margin

Competitivness 0.191*** 3.493*** 2.981*** 0.073*** 0.541*** 0.080***

(0.005) (0.143) (0.092) (0.001) (0.013) (0.002)

Competitivness*(Exporter) 0.177***

(0.006)

Competitivness*(Non Exporter) 0.172***

(0.006)

Competitivness*(Small) 0.169*** 3.039*** 1.952***

(0.006) (0.158) (0.095)

Competitivness*(Medium) 0.171*** 3.065*** 2.008***

(0.006) (0.155) (0.094)

Competitivness*(Large) 0.174*** 3.125*** 2.181***

(0.006) (0.154) (0.093)

Prob > F 0.00 0.00 0.00 0.00

Observations 23,351 16,248 26,453 26,546 26,546 26,546 23,351 23,351 16,248 26,453

R-squared 0.225 0.254 0.126 0.174 0.232 0.229 0.257 0.157


*** p

FSC is an independent, non-governmental, not for profit organization established to promote the responsible management of the world´s forests.

Printed by IT

Date post:	30-Apr-2020
Category:	Documents
Upload:	others
View:	11 times
Download:	0 times

ITC WORKING PAPER SERIES - International Trade Centre€¦ · ITC WORKING PAPER SERIES EXPLORING...

Documents