firm-level determinants of export performance: evidence ... · studies analyzing firm-level export...

Philippine Journal of DevelopmentNumber 62, First Semester 2007

Volume XXXIV, Number 1

Firm-Level Determinants of ExportPerformance: Evidence from the Philippines

MA. TERESA S. DUEÑAS-CAPARAS*

ABSTRACTThe paper determines the factors affecting the export performance offirms in three main manufacturing sectors in the Philippines. Specifi-cally, firm-level characteristics like firm size, firm age, and foreign affili-ation are identified and statistically tested to determine if they affect afirm's capability to export. The study uses a novel econometric modelthat specifically addresses fractional response behavior and estimatesthe model using a modified quasi-maximum likelihood procedure.Among the firm-level characteristics tested, foreign affiliation has themost prominent influence on a firm's propensity to export.

INTRODUCTIONThe importance of export as an economic activity and a driver of growth has longbeen established in various research endeavors. Issues addressed in these stud-ies include quantifying the contribution of export to economic growth, designingappropriate trade and industrial policies, and identifying macroeconomic factorsthat affect trade performance.

As international competition became more innovation and knowledge based,understanding trade performance went beyond the parameters of the comparativeadvantage paradigm and stressed the role of technology in affecting internationalcompetitiveness (Mytelka 2000). Focusing on the role of entrepreneurs in shaping

* Former Research Associate, Philippine Institute for Development Studies and currently Bank Officer,International Development, Bangko Sentral ng Pilipinas, A. Mabini corner P. Ocampo Streets, Malate,Manila 1004, Philippines. Email for correspondence: [email protected].

PHILIPPINE JOURNAL OF DEVELOPMENT 200788

international competition, a critical observation made is that all firms face the samemacroeconomic condition but respond and perform differently in their export ac-tivities. This suggests that there must be firm-specific characteristics that signifi-cantly influence a firm�s capability to perform in the world market. Hence, researchdirection has shifted toward understanding the different forces that influence firm-level performance. This research interest is further facilitated by the increasingavailability of large micro datasets, which unfortunately are limited to developedcountries (Wagner 2001; Sterlacchini 2001).

This paper aims to contribute to the meager but growing literature on firm-level export performance for developing countries using the Philippines as theempirical platform. As most trade-related studies in the Philippine setting are con-ducted against a macroeconomic setting, this research paper will focus on firm-level behavior and poses the question �what are the factors affecting the exportperformance of local firms?� This task is facilitated by the availability of firm-level dataset culled from the Investment Climate Survey conduced by the AsianDevelopment Bank in 2003. The paper is outlined as follows: The next sectionprovides a brief introduction and rationale of the study. Another section gives anoverview of the various studies done on the determinants of firm-level exportperformance. The succeeding section presents the theoretical underpinnings andconceptual framework of the study. The penultimate section outlines the empiricalmodel and the results. The last section gives the conclusion.

STUDIES ON EXPORT PERFORMANCE AND TECHNOLOGICALDEVELOPMENTRemarkable changes have been occurring in the global front. Focusing on manu-factured exports and technological activities, dramatic increases in both interna-tional trade and innovative behavior are evident in the sectors related to electron-ics, physics, and pharmaceuticals. For instance, using patent activity and exportshare to world trade, these sectors have shown above-average growth rates forthe period 1985�1998 (Table 1). The figures suggest that sectors that offered largetechnological opportunities exhibited the largest improvements in world exportshares. Countries like China, Malaysia, Singapore, and Thailand offered empiricalevidence supporting this observation. As these countries were advancing in tech-nological and export performance, significant structural changes also accrued intheir economies where the share of medium-high technology products to totalexports improved substantially (Montobbio and Rampa 2005).

Earlier works of economists Lindbeck and Vernon stressed the importanceof technology factors in international competition and discussed the determinantsof firm- and country-specific technological advantages. However, they differed intheir conclusions. Lindbeck espoused the view that entrepreneurial function is

DUEÑAS-CAPARAS 89Table 1. Annual growth rate of patents and export shares to world total

Annual Growth, % 1985�1998Sector Patents Exports

Computing and Data Processing 108.96 33.80Electricity and Electric Power 2.77 24.77Electronics and Components Classes 94.15 62.38Optics-radiant energy-photography 30.36 -7.37Communications and networking 94.98 14.77Electronics, Physics 42.93 25.19Pharmaceuticals 34.19 38.14Surgery-bodycare-cosmetics 62.60 8.27

Source: Montobbio and Rampa (2005).

fulfilled by talented individuals and technology needed for innovation consistingof specific knowledge of a particular product, process, and market. Vernon, on theother hand, believed that technological advantages emerge from a deliberate searchby industrial firms on knowledge needed for innovative activities that is widelyaccessible (Dosi et al. 1990). Guided by these principles and modified over time,various research works emerged that gave prominence to the role of entrepreneursin technology development. Foremost in these principles is the belief that master-ing technology is costly and firm initiated, and is in itself a source of technologicaladvantage affecting trade performance.

A study assessing the impact of technology and structural change on ex-port performance was conducted for nine developing countries, namely: Argen-tina, Brazil, China, Columbia, India, Malaysia, Mexico, Singapore, and Thailand(Montobbio and Rampa 2005). The research posited the existence of the relation-ship between technology and trade, and attempted to analyze the impact of struc-tural change on the sectoral distribution of export activities and market shares.Factors like foreign direct investment, technological specialization, skills, and re-search and development (R&D) were tested using structural decomposition analy-sis. The study confirmed that structural change is an important characteristic ofmodern economies and affects the growth trajectories of developing countries.

Sectoral and firm-specific factors affecting the trade performance of Britishmanufacturing firms were identified and analyzed (Bleaney and Wakelin 1999).Using share of export to total sales, the study concluded that higher export shareswere evident in firms that engaged in technologically innovative activities, asmeasured by their R&D expenditures. A similar study was conducted for Italianmanufacturing firms where firm size, as measured by total sales, was identified tobe the most significant factor affecting the export behavior of local firms


(Sterlacchini 2001). Firm characteristics, technological capabilities, and commer-cial capabilities were tested as possible determinants for small and medium enter-prises (SMEs) in Canada (Lefebvre and Lefebvre 2001). The study empiricallysupported the hypothesis that import activities, R&D expenditures, distributionaccess, knowledge intensity, and size significantly affect export performance. Thestudy also showed that these determinants vary according to the industrial sectorwhere the SMEs belong. Technological capabilities have the strongest effect inhigh-knowledge industries while commercial capabilities are more salient in low-and medium-knowledge sectors.

Studies analyzing firm-level export performance in developing countries wereconducted for Chile (Alvarez 2002), Mauritius (Wignaraja 2002), Ghana (Sarpongand Wolf 2004), and Indonesia (van Dijk 2002). In the Chilean firms, factors affect-ing the decision to export and the determinants of export success were identified.Productivity, firm size, and human capital were found to increase the sustainabilityof export while foreign technical licenses and foreign capital participation posi-tively improved export performance. In Ghana, the relationship between exportperformance and investment behavior of private firms was tested using a simulta-neous equation model. However, the study did not find any positive or significantrelationship between the two variables. Other factors like firm age and firm sizewere similarly tested and results indicated that younger and larger firms were likelyto invest in export compared to older and smaller firms.

In Mauritius, the export behavior of garment firms was analyzed using for-eign equity, firm size, age, technological index, and human capital (measured bythe share of engineers and technician to total employment) as possible determi-nants of export performance. Among these factors, only technological index andforeign ownership yielded significant results. A major contribution of the researchwas the construction of a technology index to capture the technological capabilityof firms. Firm-level capabilities were classified into production, investment, andlinkage activities following the taxonomy espoused by Lall (1992). A scoringsystem was assigned in each classification to build the index. A corollary objec-tive of the study was to determine the factors affecting technological capabilitiesof firms. The study showed that firm size, engineering and technical manpower,employee training, and external technical assistance have significant and positiveinfluence on export behavior while foreign ownership and age in productionshowed no significant influence. A related study stressing technological capabili-ties and firm-level export competitiveness for East and South East Asian econo-mies was also conducted focusing on three industries�electronics, auto parts,and garments (Rasiah 2003). The study showed that scale, human resource capa-bility, and institutional and systemic capabilities were some of the major factorsthat positively stimulated a firm�s export performance.

DUEÑAS-CAPARAS 91

The study of van Dijk (2002) attempted to determine the factors affectingexport performance of Indonesian manufacturing firms. It highlighted the impor-tance of sectoral variation in determining export activities and concluded thatrelative size, foreign ownership, and age were significant factors across all sectorswhile skilled labor differed according to the industry where the firm belonged. Thestudy also demonstrated that research and development activities in Indonesiaonly benefited exports in relatively mature industries while capital intensity didnot influence export behavior in scale-intensive firms.

In this study, the emerging literature on understanding export performanceis taken to the level of the firm stressing the significance of technology-relatedfactors. While it is not intended to diminish the contribution of studies that fo-cused on the macroeconomic factors, the availability of micro datasets will providean alternative way of understanding trade performance. Firm-level studies fordeveloped countries have gained momentum in the research area while those fordeveloping countries are slowing increasing. To date, there are no firm-levelstudies on export performance undertaken in the Philippines.

THEORETICAL UNDERPINNINGTrade pattern is largely explained by comparative cost advantages and relies oncost or price competitiveness. Early trade theorist David Ricardo stressed on therelative labor productivity differentials as the basis of trade and showed that eachcountry has comparative advantage�an ability to find some good it can produceat a lower relative cost, and thereby trade with other countries. This notion wasextended in the Hecksher-Ohlin model where countries were treated to have twofactors of production, labor and capital, and faced the same production functionsbut different factor endowments. The difference in relative factor endowment gen-erates trade activities. Both models treat technology either as a costless activityor irrelevant in the production and trading process. The recognition of humancapital as the �third� factor of production became the significant contribution ofthe neo-factor trade theory but still maintained the static view of the Hecksher-Ohlin model.

As the conceptualization of trade theories became more rigid, some of theassumptions adopted in the traditional trade models were relaxed like constantreturns to scale and product homogeneity. This resulted in the development of thestrategic trade models where scale economies and oligopolistic competition be-came important factors in determining trade patterns. This view was modified andembraced by the proponents of neo-technology trade theory where the emphasisis placed on the role of innovation in creating new markets and conferring costadvantages on the innovating nation. Technology became a crucial determinant ofinternational trade and differences in technologies and tastes now form the basis


of trade. Using the ideas espoused by Posner (1961) and Vernon (1966), export-ing activity is now determined by technological differences that are constantlyevolving. A major prediction of the model is that technologically advanced na-tions will export new products and import standardized ones. However, its limi-tation lies in the failure to account for a catch-up process between rich and poorcountries. Learning modes are not incorporated, resulting in the insufficienttreatment of technology.

The negligence of industry-level approach on the different market condi-tions and the capabilities of firms within the same industry became the startingpoint for the proponents of the capability framework as influenced by Austrianeconomist J. Schumpeter. Schumpeter places great value on the capitalist busi-ness enterprise and regards it as his core economic agent. He puts importance onthe monopolistic and oligopolistic market structures believing that large firmscarry out more innovative activities than small firms. He also claims the impor-tance of nonprice factors in competition, and when these factors are incorporatedin models that solely use price as a competitive leverage, the results will be signifi-cantly different. Dynamic changes are the expected outcome that will raise compe-tition to the level of new products, new sources of supply, new processes, andnew types of organization. Ultimately, it is the firm, not the industry, that decides ifit should trade or not.

The presentation of Porter�s five forces model supplements the paucity ofconcepts in understanding industry-level competitiveness (Porter 1980). Themodel developed by Michael E. Porter provides a framework for understandingcompetition in the industry level and identifies five factors that drive industrycompetitiveness. These are (i) potential entrants, (ii) buyers, (iii) substitutes,(iv) suppliers, and (v) industry rivals. By knowing these forces, the modelprovides an extensive understanding of an industry and identifies the factorsthat interrelate and influence the market participants. The model suggests that afirm wanting to gain competitive edge over its rivals needs a clear understandingof the industry where it operates. However, it is partly misleading to use anindustry-level approach in understanding export performance. It must be notedthat competitive advantage is a necessary condition to export and become aglobal market participant. However, acquiring such does not necessary lead afirm to export. Further, it has been observed that some firms perform badly inattractive industries while other firms do well in declining industries. This merelystresses the point that understanding export performance needs to be takenbeyond the industry-level approach.

The capability literature highlights the importance of the firm as the coreplayer in the acquisition and assimilation of new technology. Technical change isregarded as an activity that can be generated by firms and involves a continuous

DUEÑAS-CAPARAS 93

search and learning process that could have varying results ranging from theadaptation and improvement of chosen technology to the generation of an en-tirely new technology. These skills and knowledge are firm specific and necessaryto acquire, assimilate, adapt, change, and create technology. Enabling the firm�sperformance is a network of economic actors like other firms, suppliers, institu-tions, and the government.

DETERMINANTS OF EXPORT PERFORMANCEFirm characteristics have been identified as possible determinants of exportperformance. Some studies have shown that from these characteristics, com-petitive advantages are built and economic rents are realized. There are alsoother studies that showed that firms of the same industry differ in their perfor-mance, enactment of technology policy and corporate strategies, or use of tech-nology (Lefebvre and Lefebvre 2001). These studies simply point to one signifi-cant implication�analysis of trade performance cannot be contained to thelevel of the country or industry but has to go down the level of the firm where theimportance of firm specificity in affecting export activities cannot be neglectedor underestimated.

The relationship of firm size and export is traditionally considered as posi-tive, i.e., �to compete globally, you have to be big� (Chandler 1990 as cited byLefebvre and Lefebvre 2001). Larger firms are generally regarded as more capableof bearing the large investments and high risks associated with exporting. Severalstudies support this view empirically (Hirsch and Adar 1974, Lall and Kumar 1981,Bernard and Wagner 1996, Aitken et al. 1997, Roberts and Tybout 1997, andWignaraja and Akiara 1999 as cited by Dholakia and Kapur 1999). However, someresearches found negative or no relationship between firm size and exports. Anexplanation for this kind of relationship is the possibility that a nonlinear relation-ship exists between firm size and exports. After a certain threshold, size no longerplays a significant role. This explanation is empirically observed in Australia, Den-mark, Italy, Japan, Spain (OECD 1997 as cited by Lefebvre and Lefebvre 2001), andGermany (Wagner 2001). Another explanation offered regarding the nonlinearrelationship between size and export is that the advantages of exporting may notbe totally attractive for large domestic firms that might be oriented toward thedomestic market and that capitalize on domestic monopoly (Wakelin 1998).

Firm age and export may similarly produce conflicting relationships. As thefirm matures, it may have accumulated knowledge stock from which to build itscapabilities and provide it with better leverage to compete in the world market.However, core capabilities can become more rigid and younger firms may be moreflexible, aggressive, and proactive in catering to the demands in the world market(Lefebvre and Lefebvre 2001).


Human capital is strongly related with technological capabilities, anaspect that needs to be developed by firms to remain competitive in the exportmarket. This is usually represented by the share of the skilled employees tototal employment and/or the number of employees with degrees in either math-ematics or science. Neo-technology model suggests a strong and positiverelationship between human capital and export propensity because an edu-cated and skilled manpower possesses certain abilities that make it easier toestablish and maintain certain contacts with the foreign market. On the otherhand, the Hecksher-Ohlin model predicts that for countries with abundantunskilled labor, investment for skilled labor would be costly and will have anegative effect on exports. This is empirically tested and shown for Brazilianand Indonesian firms where human capital variable yielded statistically nega-tive relation with export performance (Willmore 1992 and Ramstetter 1999 ascited by van Dijk 2002).

Research and development expenditure is the often-used proxy variablefor technology and is expected to influence export performance positively as em-pirically tested in Brazil (Willmore 1992) and Germany (Wagner 2001). However,some studies yielded negative results between R&D and export performance spe-cifically in India (Lall 1981) and Canada (Lefebvre and Bourgault 1998). The mixedempirical results could be traced to the fact the R&D is simply a partial measure oftechnology and does not account for incremental improvements in products andprocesses. Further, the importance of R&D on export intensity differs acrosssector and country; hence, it may have a strong influence in Germany but a weakimpact in Canada.

Training of workforce is a proxy measure of technological capability and isexpected to have a positive relationship with export performance. Skill trainingenhances learning and accumulates additional skills that can improve productivityand export propensity.

Foreign interest in a local firm and export activities are expected to have apositive relationship mainly because of the multinational�s (MNE) access to supe-rior production, technology, and management know-how that the local firm canacquire. Further, MNEs have sophisticated international networks that facilitatethe exporting process. The studies on Indonesian firms validate this hypothesis(Willmore 1992 and Ramstetter 1999 as cited by van Dijk 2002).

Capital intensity is often included as a determinant of firm export perfor-mance. The Hecksher-Ohlin model predicts that capital-intensive industrializedcountries export more while the opposite is expected of labor-intensive develop-ing countries. Another explanation why a more capital-intensive firm has a higherpropensity to export is due to the past innovations and knowledge that capitalembodies, reflecting economies of scale (van Dijk 2002).

DUEÑAS-CAPARAS 95

DATA AND METHODOLOGYThe research uses data from a firm-level survey carried out by the Asian Develop-ment Bank (ADB) in collaboration with the World Bank and the Philippine NationalStatistics Office (NSO). The survey was conducted in 2002 and was primarilyintended to analyze the investment climate and productivity performance of localfirms for the period 2000�2002. A total of 716 firms were surveyed across key urbanareas using stratified simple random design. Firms belonging to these four spe-cific industries were the target of the survey: (1) food and food processing, (2)textiles, (3) garments, and (4) electronics and electrical machinery. A frequencyweight was applied, determined by the NSO, to make the survey sample represen-tative across the population size. The survey was primarily intended to obtaininformation that will help better understand and improve the investment climateand its effect on business performance. Key questions regarding the companyprofile, finance, technology, relations with other businesses, government regula-tion, contract enforcement, labor relations, and international trade were asked in atwo-part questionnaire.

For the purposes of this study, information on sales, export level, foreignequity, and the like was extracted from the survey set. Some business units, how-ever, failed to complete the survey form appropriately and had missing informa-tion. Firms for the garment and textile industries were combined into one to formthe clothing sector. Thus, firms were categorized into three major sectors, namely:food processing, clothing, and electronic. Table 2 shows the basic descriptivestatistics culled from the survey.

The following observations are drawn:Among the electronic firms, only 27 percent are nonexporters while over 50

percent are full exporters. In a reduced sampling of 97 electronic firms, 60 percent

Table 2. Descriptive statistics, ADB

Export Sale/Total Sales Food Clothing Electronics Total

(EXP)

Nonexporter 204 156 31 391where EXP=0

Partial exporter 32 48 24 104where 0<EXP<1

Full exporter 5 88 58 151where EXP=1

Missing Information 52 54 35 141Total Firms 241 292 113 646�Filtered� firms 189 238 78 505


have foreign affiliations and only 4 firms without foreign interest cater to the localmarket. The 39 firms without foreign affiliation were mainly operating to meetdomestic demand.1 The oldest electronic firm has been operating for 56 yearswhile the youngest is barely 2 years in operation.

In the clothing sector, 53 percent of the total firms surveyed are domesticallyoperating while only 30 percent are full exporters. Foreign participation in localfirms is alarming as only 52 firms of the 288 filtered firms have foreign equityparticipation of more than 50 percent. From the 52 firms, only 2 firms have noexport activities suggesting that the presence of foreign interest is vital in theexport performance of the clothing firms. There are 236 firms whose foreign equityis either zero or less than 50 percent�of these, 65 percent are nonexporters. Ma-jority of the surveyed firms have been in the business for more than 15 yearswhere the oldest clothing firm has been operating for 79 years.

The food processing sector has a reduced sampling size of 241 firms, ofwhich 204 are nonexporters. Cross tabulating for foreign affiliation, only 9 out of234 firms have foreign interest and 60 percent of these are exporters. Averagenumber of years of operation is 23 years, the longest average business lifeamong the three sectors. Table 3 presents the summary characteristics for thethree sectors.

1 Foreign affiliation is approximated by foreign equity participation of more than 50 percent.

Table 3. Mean and (standard deviation)

Firm Characteristic Electronics Clothing Food Processing

Size 0.89 0.33 0.40(0.15) (0.00) (0.01)

Proportion of 82.36 79.08 57.41skilled workforce (0.24) (0.24) (0.32)

R&D Expenditure/Sales 0.70 0.70 0.70(0.04) (0.05) (0.03)

Proportion of 48% 9% 8%exporters in theindustry providingtraining

Proportion of 58% 17% 3%exporters in theindustry withforeign affiliation

Capital Intensity 6.21 2.57 4.47(22.58) (8.32) (10.86)

Average years of 15.1 17.5 23.4operation (10.54) (13.11) (20.1)


EXPORTP2 represents the export performance of the firm defined asexport/sales. This value ranges from 0 to 1.

RSIZE represents the relative firm size defined as the number of employeesin firm j / total number of employee in sector i where firm j belongs. A larger firm,relative to the industry where it belongs, is expected to influence export perfor-mance positively since a big workforce is indicative of the firm�s capability toproduce a large level of production and meet international demand.

SKILLED represents the share of skilled workers to total workers in firm jwhere skilled workers include management and professional and skilled produc-tion workers. A large pool of skilled workers is expected to have a positive influ-ence on export performance across all sectors due to the higher labor productivityassociated with skilled manpower.

RNDSALES00 is the share of R&D expenditure to total sales in year 2000. Ahigher proportion of sales allotted for research and development activities is ex-pected to influence export performance positively. Investment in R&D is expectedto improve the knowledge and skill base of the firms and make them more capableto withstand the pressures of international competition.

TRAINING is a dummy variable where 1 represents the firm offering formaltraining to permanent employees and 0 if otherwise. This partly measures thequality of human capital where regular training of employees improves the qualityof work and consequently the firm�s export performance.

MNC is a dummy variable representing the foreign affiliation of local firms.A value of 1 is assigned if the firm has foreign equity participation of more than50 percent and 0 if otherwise. A high degree of foreign affiliation among localfirms is expected to influence export performance positively where foreign par-ticipation is regarded as an important source of knowledge and technology indeveloping countries.

CAPINTENSITY is defined as capital stock / labor cost where capital stockis the sum of value of machinery and equipment, and land, building, and leaseholdimprovement while labor cost is the sum of wages and salaries, and allowancesand bonuses. Higher capital intensity is expected to affect export performancepositively assuming that technology and knowledge favorable to local firms isembedded in the machineries and equipment.

AGE represents the years of operation. This is obtained by getting thedifference between the year 2002 and the year the firm started actual operation.A mixed effect is expected for the age variable. For the clothing and foodmanufacturing sectors, an older firm is expected to export better since it takesyears to learn about the market it caters whereas for the electronic firms, ayoung firm may be able to penetrate the export market as effectively as theolder firm due to the fast pace of technology. A younger firm may enjoy


procedure is not appropriate for this type of estimation due to its failure to recog-nize the restrictions in the data definition (bound by 0 and 1) while the Tobitestimation is used for censored variables. The ADB firm-level dataset providesfull information (0 < export performance < 1) on the export propensity of firms andcensoring the dataset (i.e, filter export performance = 0) will yield biased results.The same estimation procedure was used by van Dijk (2002) in determining theexport performance of Indonesian firms. The Papke-Wooldridge model can beestimated using the statistical package STATA 9.0 under generalized linear modelwith Logit as the link function and robust estimators.2

Results of the econometric test are shown in Tables 5, 6, and 7. The empiricalmodel attempts to determine if characteristics like firm size, skills, training, R&Dexpenditure, foreign affiliation, capital intensity, and age can influence the exportperformance of firms in selected Philippine manufacturing sectors. Regressionanalyses were conducted on an industry-level basis stressing the significance ofsectoral variation. This means that for the same firm characteristic, it is possiblethat it may have a different influence on export behavior depending on whichsector the firms belongs. Pooled regression analysis is not conducted to empha-size sectoral variation.

Food Processing SectorUsing the general model where all the possible determinants were included in theestimation, only SKILLED and MNC showed positive and significant influenceof export performance as expected. Foreign affiliation appeared to be the pri-mary source of knowledge and technology for local firms while technical skillsimproved the quality of production, thereby making the goods more attractive inthe world market. Size and export performance exhibited an inverted U relationbut the z-tests were insignificant rendering the nonlinear effects of size incon-clusive. Similarly, RNDSALES00, AGE, and CAPINTENSITY variables yieldedsigns contrary to expectations but were not significant.

An alternative or reduced-form model was estimated by removingRNDSALES00 from the original function. The rationale behind the alternative orreduced model is the belief that most local firms hardly do R&D activities. Theresults were similar to the general model but with a slight increase in the level ofsignificance for the SKILLED and MNC variables. Obtaining the marginal effects,the results indicate that by increasing the proportion of skilled workers in the workforce by 1 percent, export performance will increase by 12 percent assuming all

2 See Wagner (2001) for the details of the estimation procedure and Papke and Wooldridge (1996) for thefractional response modeling.

DUEÑAS-CAPARAS 101Table 5. Estimation results for food processing sector

Variable Estimation 1 Estimation 2 MarginalGeneral Model Alternate Model Effects

Constant -4.70 -5.04(-5.56) (-6.39)

Rsize 18.06 4.73 0.165(0.25) (0.06)

(Rsize)2 -147.52 -181.70 -6.329(-0.32) (-0.15)

Skilled 3.42* 3.50* 0.122(2.20) (2.31)

RNDSales00 -54.77(-1.14)

Training 1.11 0.76 0.034(1.24) (0.90)

MNC 1.24* 1.64* 0.123(2.10) (2.71)

Capintensity -0.00 0.00 0.000(-0.29) (0.26)

Age -0.56 -0.04 -0.001(-1.15) (-0.87)

(Age)2 0.00 0.00 0.000(1.73) (1.23)

AIC 1.82 1.81No. of firms 189 189

Note: z values are reported in parenthesis; * means significant at 95 percent interval.

things are constant. Similarly, a 1 percent increase in foreign equity participationwill improve the proportion of export to sales by 12 percent.

Clothing SectorVariables MNC, RSIZE, and AGE were significant with the expected signs. Thismeans foreign interest in the clothing firms is an important determinant of exportperformance and firm maturity matters suggestive of scale economies and vintageeffects. The square for AGE and RSIZE were included to test for possible nonlin-ear effects of firm age and firm size. The z-test turned out significant with anegative sign for both squared variables. This means that the effect of age andsize can be depicted by an inverted U-shape implying that at a certain thresholdlevel, the positive effect of firm maturity and size will diminish. It implies that theadvantages of size and age can only be reaped at a certain point and beyond that,further expansion will no longer be profitable.


SKILLED, TRAINING and CAPINTENSITY did not yield significant re-sults while RNDSALES00 was significant but with the opposite expected sign.The result of the latter variable is similar to the one conducted by Lall (1981) forIndian engineering firms. Possible explanation offered for the significant yetnegative effect of R&D is that the variable is only a partial measure of technol-ogy and does not take into account incremental improvements in processes thatthe industry is more prone to experience. This is generally the case for develop-ing countries where R&D is low due to the adaptive nature of technical change(van Dijk 2002). In the Philippine setting, a possible explanation is that clothingfirms source their technology and knowledge from the capital goods they importand from suppliers of materials. Investment in R&D activities would likely bur-den their core activity given limited financial resources.

Table 6. Estimation results for clothing sector


Constant -1.82 -1.84(-3.02) (-3.15)

Rsize 295.29* 232.09* 53.93(2.55) (2.65)

(Rsize)2 -9885.13* -6162.76* -1431.91(-2.46) (-2.66)

Skilled -0.34 -0.17 -0.041(-0.57) (-0.29)

RNDSales00 -10.11*(-1.88)

Training 0.47 0.38 0.091(0.71) (0.64)

MNC 5.81* 5.01* 0.741(4.98) (5.60)

Capintensity -0.00 -0.00 -0.000(-0.02) (-0.08)

Age 0.08* 0.07* 0.015(1.80) (1.80)

(Age)2 -0.00* -0.00* -0.000(-1.88) (-2.01)

AIC 4.06 4.04No. of Firms 238 265

Note: z values are reported in parenthesis; * means significant at 90 percent interval.

DUEÑAS-CAPARAS 103

Electronics SectorVariables RNDSales00, TRAINING, MNC, and CAPINTENSITY yielded

highly significant results with the expected signs. The study empirically confirmsthat a local firm that invests in R&D, conducts regular skill training to its employ-ees, has considerable foreign influence, and has a high capital per employee ratiowill have a higher propensity to export. A higher R&D expenditure to sale ratiosuggests that firms should reinvest in learning to improve their performance. Pro-viding complementary training to employees will further boost the productivityperformance of firms. Foreign affiliation becomes the local firm�s primary source ofknowledge and network advantage. This knowledge source is deepened with theaccumulation of capital goods where technology is likely embodied. The statisti-cal results also partly confirm capital-labor complementarity where higher capitalintensity requires better-trained employees.

Table 7. Estimation results for electronics sector


Constant -8.31 -5.05(-3.70) (-2.79)

Rsize 372.80 304.61 65.63(1.68) (1.41)

(Rsize)2 -3734.93 -3010.30 -657.58(-1.43) (-1.18)

Skilled 1.29 1.54 0.23(1.16) (1.16)

RNDSales00 275.56* 48.52(3.43)

Training 2.77* 1.99* 0.48(3.31) (2.75)

MNC 2.94* 2.43* 0.52(4.06) (2.95)

Capintensity 0.24* 0.13* 0.04(3.05) (2.29)

Age 0.91* 0.22 0.07(2.12) (1.15)

(Age)2 -0.01* -0.01 -0.00(-2.65) (-1.83)

AIC 1.52 1.79No. of Firms 78 95

Note: z values are reported in parenthesis; * means significant at 95% interval.


Variables RSIZE, AGE, and SKILLED have the expected signs but were notstatistically significant. The nonlinear effect of size was also incorporated but didnot yield significant result. The variable age-squared (age)2 was included andturned out significant. This signifies that as the firm ages, the effects of cumula-tive learning and training improve firm performance but after a certain age thresh-old, the returns decline, suggesting that firm experience only matters at a certainpoint. Considering the rapid pace of technical change occurring in the sector, thisis not a surprising result since firms must continually adapt to these changes inorder to grasp the profit opportunities presented in the world market.

SUMMARY AND CONCLUSIONThe study analyzed the different factors that could affect the export performanceof firms in selected Philippine manufacturing industries. The factors identified arebasically firm specific like size, percentage of skilled labor to total labor, trainingactivities, foreign affiliation, R&D activities, capital intensity, and firm age. Thestudy used the information-rich survey conducted by the ADB in 2002. The clas-sification of the selected manufacturing sectors stressed the importance of sectoralvariation in determining the influence of these firm-specific factors to export per-formance. The possible relation between export performance and firm-level char-acteristics was tested using a novel econometric model by Papke and Wooldridgethat was specifically developed for fractional response modeling.

The main findings of the study are summarized as follows:✦ The influence of foreign affiliation is similar across all sectors�posi-

tive and strongly influential in improving a firm�s export performance.The variable MNC is the only factor that is tested statistically signifi-cant in all the three major sectors with the expected signs. The strongpositive influence of MNC on a firm�s export behavior confirms thebeneficial effects of foreign participation in locally initiated endeavors.

✦ Research and development activity is highly influential for science-based firms and confirms the belief that it is a necessary ingredient thatwill propel export propensity.

✦ Development of human capital through training strongly influencesthe export performance of science-based sectors. This magnifies theimportance of learning by training in improving the performance andproductivity of firms (Bell 1984). The effect of training on clothing andfood processing firms, together with the share of skilled workers acrossall sectors, is not conclusive.

✦ A higher capital per worker positively influences the export perfor-mance of electronic firms but not for the clothing and food processingsectors. This is reflective of the capital-intensive nature of the sector

DUEÑAS-CAPARAS 105

and the possibility of knowledge and technology embodied in the capi-tal goods used by the electronic firms.

✦ Firm size across all sectors exhibits an inverted U-shape suggesting anonlinear relation between size and export performance. This relation-ship is most significant for the clothing sector. An inverted U-shaperelation means that as firms expand their operation, as measured by thenumber of employees relative to the industry, this move will have abeneficial effect on their export activities. However, after a certain levelof expansion, any increases in the firm size will have a less-than-pre-ferred outcome in export performance.

✦ Firm age is an important factor in the export performance of electron-ics and clothing sectors. For the clothing sector, a firm�s length ofoperation is suggestive of gains in scale economies and beneficialeffects of deep knowledge in its customer base. For the electronicfirms, however, maturity in operations will only be beneficial at acertain point. Beyond the threshold age, experience does not matter.This is a possible outcome of rapid technological changes in thesector where a younger and more flexible firm may be able to addressimmediately the changing consumer taste and preference for techno-logical advancement.

In terms of policy initiatives, the importance of local firms to be affiliatedwith the foreign firms should be stressed. Be it in the form of foreign equityinfusion, joint ventures, licensing agreements, or direct investment, foreign firmscarry strong network linkages with the international community that may be ben-eficial to the local firms.

Investments in R&D activities and training of employees significantly im-prove the export propensity of electronic firms. The merits of R&D activities havelong been established in various studies in spite of the existence of some studiesthat say otherwise. It is stressed, however, that R&D activities are not limited toinnovative activities but can be applied to improving technological capabilitiesthat are necessary for firms in developing countries.

Intensive acquisition of capital goods relative to increases in employment isa necessary ingredient for export performance in science-based sectors. With theknowledge and technology embodied in these capital goods, local firms will ben-efit through improvement in productivity and efficiency.


REFERENCESAlvarez, R. E. 2002. Determinants of firm export performance in a less developed

country. California: Anderson Graduate School of Management, UCLA.Bell, M. 1984. Learning the accumulation of industrial technological capacity in

developing countries. In Technological capability in the Third World,edited by M. Fransman and K. King. London: MacMillan

Bleaney, M. and K. Wakelin. 1999. Sectoral and firm-specific determnants of exportperformance: evidence from the United Kingdom. Centre for Research onGlobalization and Labor Market Research Paper 99/12. United Kingdom:School of Economics, University of Nottingham.

Dosi, G., K. Pavitt and L. Soete. 1990. The economics of technical change andinternational trade. New York: Harvester Wheatsheaf.

Dholakia, R.and D. Kapur. 1999. Determinants of export performance of Indianfirms � a strategic perspective [online]. htt://www.iimahd.ernet.in/publi-cations/data/2004-08-01.

Lall, S. 2000. Export performance and competitiveness in the Philippines. QueenElizabeth House. Working Paper No. 49. United Kingdom: Universityof Oxford.

��. 1992. Technological capabilities and industrialization. World Develop-ment 20(2):165�186.

Lall, S. and R. Kumar. 1981. Firm-level export performance in an inward lookingeconomy: the Indian engineering industry. World Development9(5):453�463.

Lefebvre, E. and L. Lefebrve. 2001. Innovative capabilities as determinants ofexport performance and behavior: a longitudinal study of manufacturingSMEs. In Innovation and firm performance: econometric explorationsof survey data, edited by A. Kleinknecht and P. Mohnen. London:MacMillan and Basingstoke: Palgrave.

Lefebvre, E. and M. Bourgault. 1998. R&D-related capabilities as determinants ofexport performance. Small Business Economics 10:365�377.

Montobbio, Fabio and F. Rampa. 2005. The impact of technology and structuralchange in export performance in nine developing countries. World Devel-opment 33(4):527�547.

Mytelka, L. 2000. Local systems of innovation in a globalized world economy.Industry and Innovation 7(1):15�32.

Papke, L.E. and J.Wooldridge. 1996. Econometric methods for fractional responsevariable with an application to 401 (K) plan participation rates. Journal ofApplied Econometrics 11(6):619�632.

Porter, M.E. 1980. Competitive strategy: techniques for analyzing industries andcompetitors. New York: Free Press.

DUEÑAS-CAPARAS 107

Posner, M.V. 1961. International trade and technical change. Oxford EconomicPapers 13: 323�341.

Rasiah, R. 2003. Export experience and technological capabilities: evidence fromEast and Southeast Asian firms. Manila: Asian Development Bank.

Sarpong, D. and S. Wolf. 2004. Export performance and investment behaviorof firms in Ghana. Paper presented at the ISSER/Cornell UniversityConference on Ghana�s Economy at the Half Century.

Sterlacchini, A. 2001. The determinants of export performance: a firm-level studyof Italian manufacturing. Weltwirschaftliches Archive 137:450�472.

Van Dijk, M. 2002. The determinants of export performance in developing coun-tries: the case of Indonesian manufacturing. Working Paper No. 02.02.The Netherlands: Eindhoven Centre for Innovation Studies,

Vernon, R. 1966. International investment and international trade in the productcycle. Quarterly Journal of Economics 80:190�207.

Wagner, J. 2001. A note on firm size�export relationship. Small Business Econom-ics 17:229�237.

Wakelin, K. 1998. Innovation and export behavior at firm level. Research Policy26(7):829�841.

Wignaraja, G. 2002. Firm size, technological capabilities and market-oriented poli-cies in Mauritius. Oxford Development Studies 30(1):87�104

��. 1998. Trade liberalization in Sri Lanka: exports, technology and industrialpolicy. London: Macmillan Press.

Wilmore, L. 1992. Transnational and foreign trade: evidence from Brazil. Journal ofDevelopment Studies 28 (2):314�335.

firm-level determinants of export performance: evidence ... · studies analyzing firm-level export...

Documents