Sectoral Differences in Determinants of Export Intensity

This study investigates firm characteristic determinants of export intensity in small firms. The originality of our approach is a comparative analysis of export intensity between firms in the computer software and manufacturing sectors, using a quasi-maximum likelihood estimation to test for the correct specification of the conditional mean model. Results indicate that larger, younger firms have greater export intensity in the computer software sector than in manufacturing. Research and development expenditure is equally important for export intensity in both sectors, but patent income is not significant. Sourcing managerial advice and expertise from the national development agency is important for firms in the manufacturing industry, but not for computer software firms. It is therefore important for export promotion organisations to publicise supports, as few small firms are aware of their availability. Our findings are especially valuable for policy makers concerned with low levels of export intensity among small firms.


Introduction
Establishing a presence in international markets through exporting goods and services is especially important for the growth and sustained development of small and medium sized enterprises (SMEs) (Westhead et al., 2001). However, despite the fact that SMEs account for over 60% of private sector employment and contribute at least half of the Gross Value Added (GVA) in many developed and developing countries, only a small proportion of SMEs sell goods and services in foreign markets (Bell, 1997;Westhead et al., 2002Westhead et al., , 2004. 25% of EU27 SMEs are exporters (European Commission, 2010) and less than 5% of US SMEs (USITC, 2010). Addressing this dearth of SME export orientation has become a priority for policy makers across the globe, as witnessed by increased efforts to boost exports through export promotion organisations (EPOs) (Lederman et al., 2010).
Research on SME export determinants and intensity have surged in the past two decades, although these studies have concentrated primarily on the manufacturing sector, with empirical studies typically comprising of comparative studies between manufacturing activities (e.g. Wagner, 2001) and cross-country differences between manufacturing plants (e.g. Roper and Love, 2002). There are relatively few investigations of the export determinants or intensity of services firms (Sousa et al., 2008). This is a considerable omission, given the phenomenal growth in services exports over this period (OECD, 2011). In Ireland, for example, although the real value of manufacturing exports has remained relatively static since 2000, the value of services exports has risen by 322% from €18 billion to €74 billion, and accounts for 48% of exports (Forfás, 2011).
Services firms differ significantly from the manufacturing sector, particularly in terms of age, size, and differences in innovation behaviour (Pires et al., 2008). Additionally, there are significant differences within the services sector between large scale services such as banking, knowledge intensive services, and smaller scale services (Audretsch et al., 2004). There are also important variations between services firms in innovation and technological change (Evangelista, 2000;Miles, 2005). These differences have important implications for exporting and export intensity, and are of particular concern for policy makers seeking to increase export capacity through provision of supports.
This paper aims to join to a number of commendable recent studies, such as Wagner (2011) which have begun the task of addressing this research lacuna. Specifically, we employ firm level data to investigate sectoral differences in export intensity between SMEs, comparing the manufacturing and computer services sectors.
We study this issue through a quantitative analysis of the export performance of Irish SMEs. Given that small firms face many disadvantages in competing in international markets (Alvarez, 2004), due primarily to economies of scale and access to resources (Wagner, 2001), policy makers in many countries expend considerable resources addressing these issues though the establishment of EPOs and other public support programmes. Ireland is no different in this respect. Data for this study was sourced from a survey of 702 Irish SMEs, consisting of firms with between 20 and 250 employees. This sample size is larger than that of previous studies of export determinants and intensity, for which Sousa et al. (2008) cite a mean of 260 firms. The quantitative methodology employed in this study is the recently developed one-stage fractional probit technique of Ramalho et al. (2011). The benefits of this particular technique are discussed in detail in Section 3.
The research question addressed in this study is: Are firm characteristic determinants of export intensity in the manufacturing sector different from those in the computer software and services sector? We also assess whether the impact of export promotion agencies on export determinants and intensity differs across sectors. The implication of this research is that, because firm characteristic determinants of export intensity differ between sectors, policy makers seeking to support and promote export activity should design and provide supports geared towards each sector (and not a one size fits all approach). Our contribution to the literature is thus twofold, as we (a) identify sectoral differences in export intensity, and (b) suggest how EPOs could better target export supports to small firms.
The rest of the paper is structured as follows: in the following section we review the literature and formulate hypotheses. The background to the study and methods of data collection are explained in section 2, and the research methodology is described in section 3. Results are presented and discussed in section 4, and conclusions and policy implications are outlined in the final section.

Theoretical framework and derivation of hypotheses
Internationalisation research has developed significantly from the earliest studies which concentrated predominantly on multinational firms. Subsequent approaches placed increased emphasis on small and medium sized enterprises (SMEs), and considered a number of stage-model approaches, commencing with the Uppsala internationalisation model (Johanson and Vahlne, 1977). Further theoretical developments went beyond the stage-model approach, which was considered inadequate to explain phenomena such as the emergence of 'born global' firms (Zahra, 2005) (see Wright et al., 2007 for a critique). Academic studies may be broadly categorised in two distinct but not unrelated strands grounded in 'the entrepreneurship literature' and economics literature respectively. 3 The former concerns the process of internationalisation, and it emerged from studies investigating SMEs seeking to export shortly after establishment. These 'International New Ventures' (Oviatt and McDougall, 1994) did not proceed through a number of stages as theorised by previous approaches, but endeavoured to "…derive competitive advantage from the use of resources and the sale of outputs in multiple countries (from inception)…" (Oviatt and McDougall, 1994, p.49). This literature draws from a number of different theories, particularly the resource based view of the firm, and empirical studies employing this approach examine personal characteristics of the firm owner, including management and industry knowledge (Westhead et al., 2001), intention (Kunda and Katz, 2003) and networks and alliances (Zain and Ng,3 Notwithstanding the application of a number of a number of different theories to the subject, Sousa et al. (2008) state that the literature is 'fragmented and atheoretic '. 2006). Highlighting the importance of the resource-based view, these papers emphasise that variables influencing export intensity in SMEs and large firms are significantly different.
Internationalisation studies emanating from the microeconomics literature focus on firm characteristics as determinants of export activity and intensity, along with the effect of exporting on innovation and performance. Evidence suggests that there are a number of important firm characteristics which may influence the propensity of a firm to export, and in turn, the export intensity of that firm. Understanding the determinants of export intensity is important because, similar to employment growth, only a handful of firms account for the vast bulk of exports. These studies focus predominantly on manufacturing firms, and commonly investigated firm characteristics include firm size and age, expenditure on research and development, and productivity. Researchers increasingly focus on the services sector, employing a broadly similar approach.
Notwithstanding the importance of both sectors to national export capacity, there is a lack of comparative research on the most export intensive sectors.
We adopt the latter approach in conducting this comparative study of export intensity.
We are thus examining potential determinants of export intensity, rather than the process of internationalisation. Although these approaches are not mutually exclusive, our study is couched in this microeconomic literature. Crucially, we want to consider firm characteristics for both manufacturing and services firms, and to explore whether these characteristics have different impacts. We now look in more detail at the firm characteristics identified in this literature, in formulating hypotheses which we will test for both the services and manufacturing sectors.
Larger firms have more resources with which to enter foreign markets (Roper and Love, 2002), and have greater capacity to overcome sunk costs associated with foreign market entry (Ottaviano and Volpe Martincus, 2011) such as information gathering or economies of production and marketing (Wakelin, 1998). This is even more relevant in the case of SMEs (Ottaviano and Volpe Martincus, 2011). Larger firms also have more opportunities to raise finance, and are expected to have more technological resources available (Harris and Li, 2008). The size effect is typically non-linear (Wagner, 1995;Wakelin, 1998;Sterlacchini, 1999;Bernard and Jensen, 1999), and has been characterised as an inverted u-shape (Gourlay et al., 2005), due to disincentives to export arising from a large presence in the domestic market (Wakelin, 1998) or increased co-ordination costs as the scale of the operation increases (Wagner, 2001).
Other studies report a u-shaped relationship (Chiru, 2007) and no significant effect (Love and Mansury, 2009).
Self-selection of larger, more productive firms may be less prevalent among services exporters than their manufacturing counterparts. Additionally, Eickelpasch and Vogel (2009) note that capital intensity as an indicator of firm assets, embodying past innovations and capturing economies of scale, is expected to have a positive effect.
Services firms, particularly knowledge-based ones, may be relatively less capital intensive than manufacturing firms.

H1: Export intensity is positively related with firm size
H1a: Size has a greater effect on the export intensity of manufacturing firms than 'computer software and services' firms, ceteris paribus. Love and Mansury (2009) note that there is no consensus regarding the role of the age of a firm on export propensity. On one hand, older firms may have had more time to establish and expand distribution networks, as well as gain a market position in export markets. On the other hand, older firms may experience inertia and inflexibility in the face of changing market conditions (Contractor et al., 2007). Roper (2006) finds high export propensity among younger Irish manufacturing firms, though Majocchi et al (2005) report that age is positively associated with export intensity for Italian SMEs.
Eickelpasch and Vogel (2009) also point to the incremental process of internationalisation, where firms first enter markets that are similar to their home market, as well as the importance of internal firm resources such as management strategies and characteristics as potential determinants of a firm's export performance.
In light of this inconclusive evidence, we tentatively propose the hypothesis:

H2: Export intensity is negatively related with firm age
Firms in the manufacturing sector are generally older than firms in the services sector (Berggren et al., 2000). Additionally, the 'age effect' for manufacturing firms may be greater than services firms due to time required for product development, and establishment of distribution networks. Therefore, we hypothesise that: H2a: The negative relationship between export intensity and firm age is of greater magnitude for firms in the 'computer software and services' sector than manufacturing firms.
Early studies investigating the effect of innovation on exporting at the firm level typically use the level of R&D expenditure as a proxy for innovation. A more nuanced approach is taken in recent studies, and a number of measures of innovation are now employed (Lefebvre et al, 1998). We follow this recommendation by employing two measures of innovationan input variable (R&D expenditure) and an output variable (patent income). Empirical studies find that innovation (as measured by internal R&D expenditure or innovative products) has a positive effect on exports, both in manufacturing (Wakelin 1998;Sterlacchini, 2001;Wagner, 2001;Roper and Love, 2002,), and services (Gourlay et al., 2005;Chiru, 2007;Love and Mansury, 2009).
Firms that invest in product improvement (Ottaviano and Volpe Martincus, 2011), and invest in internal research and development (Ganotakis and Love, 2011) will, consistent with the resource based view, have a competitive advantage over their competitors and be more likely to enter foreign markets. Similar to Love (2002, p.1093), we argue that R&D expenditure is not regarded as an indicator of innovation, but as "…an indicator of investment in the resource base of the plant".
Thus, it is proposed that firms with a greater investment in unique and non-imitable assets have greater export intensity. (It could also be stated that firms willing to make a greater investment in internal R&D have greater capacity, are generally larger (Wakelin, 1998;Gourlay et al., 2005), and consequently have a greater need to export):

H3: Export intensity is positively related with expenditure on research and development
For firms in sectors with a greater emphasis on technological innovation, opportunities and product cycle differences, the sectoral effect is expected to be significant in conditioning firm's export-innovation relationship (Harris and Li, 2009). Anderton (1999) and Ioannidis and Schreyer (1997) find that R&D expenditure and patenting activity are more important in technology intensive industries. Innovation has been found to positively influence the probability of exporting in business services (Ebling and Janz, 1999;Gourlay et al., 2005;Love, Roper and Hewitt-Dundas, 2009;Love and Mansury, 2009) and manufacturing sectors (Wagner, 2001;Roper and Love, 2002).
Considering this evidence, we do not propose that there is a significant difference between the effects of R&D expenditure on export intensity between the 'computer software and services' and manufacturing sectors. Roper and Love (2002) state that differences in the effect of R&D spending on levels of exporting intensity in previous studies indicate that reliance on this measure alone may be unreliable, and advise using a range of innovation indicators. Anón Higón and Driffield (2011, p.6) highlight the need to "…measure innovation more carefully than simply through R&D spend…". A number of studies have modelled the propensity to innovate employing a lagged variable, or employing the innovation history of firms.
For example, Wakelin (1998) finds that the number of past innovations is positively related with the probability of an innovative firm exporting. There are a number of indicators of past innovation, not least of which is whether firms have created and developed income-generating patents. Consistent with evidence from previous studies, we hypothesise that firms with greater levels of past innovation in the form of incomeproducing patents have a greater intensity of exports:

H4: Export intensity is positively related with income from patents
A number of studies investigate the role of export promotion organisations in the internationalisation of new firms (e.g. O' Gorman and Evers, 2011), and in deploying export promotion instruments (Alvarez, 2004;Yannopoulos, 2010). These studies highlight the importance of government agencies in supporting SMEs exporting to new markets, especially mediation and information gathering, identifying opportunities and potential customers, and developing and expanding export capacity. Although it may seem tautological to propose a positive relationship between export intensity and receipt of support from a government development agency, the variable we employ to test the latter is 'receipt of managerial advice and expertise' from the national agency for enterprise development, Enterprise Ireland. This advice is not specifically related to exporting per se, but is more a measure of 'outward orientation', as SMEs rely primarily on internal resources for advice and expertise. Therefore, we hypothesise: H5: Export intensity is positively related with receipt of managerial advice from a national development agency.
It is not apparent whether receipt of managerial advice and expertise from the national enterprise agency has a proportionately greater effect on export intensity in either sector. It may be argued that this expertise has a larger effect for manufacturing firms because, as they are older, there is a greater likelihood that they will have approached the national development agency for advice and assistance (as supports for manufacturing industry have been established longer). On the other hand, Barry and Van Egeraat (2008) attribute the stellar growth of the indigenous software sector to the intensive supportive role played by Enterprise Ireland. On the balance of evidence we propose that: H5a: The positive relationship between export intensity and receipt of advice from a national development agency is of greater magnitude for manufacturing firms than firms in the 'computer software and services' sector.
The most important firm characteristics highlighted in the literature are included in our hypotheses, although other microeconomic studies of this type examine employee wages (Schank et al., 2007), labour productivity (Wagner, 2011), sources of information (Lefebvre et al., 1998), owner characteristics (Roper and Hewitt-Dundas, 2001), and employee training activities (Ottaviano and Volpe Martincus, 2011).

Background and data collection
In common with other small, open economies, the Irish economy is highly internationalised, as the value of exports and imports amount to 137% and 103% of Gross National Product respectively (Central Statistics Office, 2012, 2013a. There has been a sizeable growth in the volume of exports and imports to Ireland in the past 20 years (Forfás, 2011). Although foreign owned large multinationals produce the bulk of services and manufacturing exports, Irish owned SMEs produce 7% of total direct exports in these sectors, which amounts to 7% of GVA (Lawless et al, 2012). The importance of indigenous exporting SMEs for employment is even more significant, as they account for 23.5% of employment in the manufacturing sector, which is double that of foreign exporting SMEs (Lawless et al., 2012). Similarly in the services sector, total employment in indigenous exporting SMEs is more than twice that of their foreign counterparts (ibid). Internationalisation has been established relatively longer in the manufacturing sector, as the Irish industrial landscape and trade statistics are significantly influenced by an economic development policy of pursuing Foreign Direct Investment (FDI), which initially focussed on attracting manufacturing plants (Ó Gráda, 1997). In recent decades, investment from international services firms is increasing in importance, and this sector now accounts for over 50% of total exports (Central Statistics Office, 2013a). Whilst this growth can be largely attributed to foreign multinationals, there have been important overflows to indigenous entrepreneurship, particularly in the computer software sector (Acs et al., 2007). The computer software and services sector is one of the three primary sectors driving the growth in exports, as 7 of the top 20 exporting firms are in this sector (Irish Exporters Association, 2011). Computer services exports have grown from €6 bn in 2000 to €32 bn in 2011, representing over 40% of services exports (ibid). Whilst a large proportion of these exports are accounted for by multinational firms, a substantial indigenous industry has emerged in parallel (Barry and Van Egeraat, 2008).
Data for this study was sourced from a survey of SMEs in the Republic of Ireland, consisting of firms with between 20 and 250 employees. The original database of 1,502 firms was substantially cleaned to remove non-independent enterprises, along with companies in the financial sector. The questionnaire instrument was distributed to the remaining 702 firms using a multimodal approach. This methodology yielded 299 responses, representing a response rate of over 42%. Information collected includes firms' exporting and innovation activity, along with details on sources of managerial advice and expertise, and financial resourcing. A detailed profile of respondent firms is provided in table 1. Exporters account for almost three quarters of the sample, which is significantly more than that recorded in previous studies (e.g. Eickelpasch and Vogel, 2009;Ottaviano and Volpe Martincus, 2011). A quarter of the sample report that revenue from exports account for less than 10% of total revenue. Firms in the manufacturing and 'computer software development and services' sectors have a significantly greater proportion of export revenue than firms in the 'distribution, retail, hotels and catering', 'other Section A.
Section B. Section C.

Research Methodology
There has been a significant shift in the methodological approach applied in studies investigating export determination and intensity. The standard methodology employed in earlier studies follows a two-step approach, which consists of first modelling the decision to export, employing a binary dependent variable (exporter/non-exporter). This is commonly modelled as a probit regression. The second step involves modelling the 'decision' of export intensity, which is conducted by estimating another model (commonly by comparing an unrestricted model against a restricted model, frequently a Tobit regression). Examples of these studies include Wakelin (1998), Sterlacchini (1999, Roper and Love (2002), and Gourlay et al. (2005). Each study rejects the restriction implicit in the Tobit estimator, with the exception of Sterlacchini (1999) who reports varying acceptance of single-censored Tobit model. These studies typically use identical independent variables for both steps. Wagner (2001) states that the two-step approach is imperfect, as exporting is "…not a two-step decisionto export or not, and then how much to export" (Wagner, 2001, p.230). He states that it is consequently difficult to identify a distinct set of variables appropriate for modelling both aspects of the two-step approach separately. An alternative to the commonly applied two-step approach is the one-step approach, in which all observations (both exporters and non-exporters) are included in estimating the model. Ordinary Least Squares (OLS) is not appropriate as the dependent variable is confined between zero and one and a large proportion of observations reside at the zero boundary (Wakelin, 1998;Wagner, 2001). Some studies apply a Tobit regression (e.g. Lefebvre et al., 1998), although "…the censored regression model (tobit) is only applicable in cases where the latent variable can take negative values, and the observed zero values are a consequence of censoring and non observability" (Maddala and Lahiri, 2009, p.347). This does not hold for the dependent variable, however, as it is not censored, nor can exporting take negative values. Another approach adopted in the literature is to model the data as a beta distribution. The beta regression is also inappropriate, as it ignores observations at the extremes of the distribution. This is an important limitation as a large number of firms do not export at all (Wakelin, 1998), and "…observations at the boundaries of a fractional variable are a natural consequence of individual choices and not of any type of censoring …" (Ramalho et al., 2011, p.22). For these reasons we concur with Wagner (2001) that exporting and export intensity is not a two-step process, and we adopt a one-step approach.
A number of authors state that linear models are inappropriate when investigating how exogenous variables influence a fractional response variable (e.g. Ramalho et al., 2011 "…the signs and significance levels obtained using fractional response models are very similar to those obtained using Tobit". The dependent variable in the present study is fractional, and was collected in interval form. We select the mid-point of each interval in running the fractional response models. Similar to Ramalho et al. (2011), we consider only quasi-maximum likelihood (QML) estimation as it outperforms all non-linear least squares (NLS) estimators. As the fractional response variable is not continuous, we also run a number of interval regression models and an Oprobit model, results of which are reported in appendix 1 (table 6). The signs and significance of the variables in all methods are the same, and regression coefficients are broadly similar. Hypotheses formulated in section 2 were empirically tested by calculating QML regression models as proposed in the full testing methodology of Ramalho et al. (2011). The basic model tested is represented by: Y = β 0 + β 1 SIZE + β 2 AGE + β 3 R&DEXP + β 4 PATENT + β 4 EIADVICE + ε Additional models were estimated to test for inter-industry differences employing dummy variables. Y = β 0 + β 1 SIZE + β 2 AGE + β 3 R&DEXP + β 4 PATENT + β 4 EIADVICE + β 7 METAL + β 8 MFCT + β 9 SERVS + β 10 COMPUTER + β 11 OTHER + ε Following Ramalho et al. (2011), Logit, Probit, Loglog and Cloglog models were estimated. These nonlinear models use the logistic, standard normal, extreme maximum, and extreme minimum distribution functions respectively.

Results
Results for four specifications of one-part fractional regression models for the total sample are presented in appendix 1 (  contrary to Sterlacchini (2001). By contrast, the positive relationship between innovation outputs (patent income) and export intensity is insignificant for all models apart from the Cloglog specification. This evidence provides support for hypothesis 3, but hypothesis 4 is rejected. Export intensity is also positively related with receiving managerial advice and expertise from Enterprise Ireland, the national government agency for supporting Irish businesses in the manufacturing and internationally traded services sectors. Hypothesis 5 is therefore accepted, although these firms may be selfselecting. In summary, firms with a greater export intensity are larger, invest more in innovative activities, and are more 'outward looking' in seeking managerial advice and expertise from the national development agency.
A preliminary investigation of sectoral differences was conducted using the dummy variable approach. Results indicate that firms in the internationally traded sectors, i.e.
manufacturing and knowledge intensive sectors (computer software development and services), have a greater export intensity than firms in the reference sector, 'distribution, retail, hotels and catering'. This result holds for all models. Firms in the 'other' sector also have a higher intensity of exporting, but this result is not significant for all specifications (specifically for the probit, loglog, and interval regression models). Results for firms in the 'other services' sector, which are predominantly focussed on the local market, are negative and insignificant.
A more detailed examination of sectoral differences was conducted by estimating the basic regression specification separately for the two largest exporting sectors, the manufacturing and the 'computer software development and services' sector. Results presented in tables 3 and 4 indicate that although the direction of most coefficients is similar, and the same as models for the total sample, there are differences in the size and significance of coefficients between the two sectors.
The effect of firm size on export intensity is greater for firms in the 'computer software development and services' sector than in manufacturing, leading us to reject hypothesis 1a. Coefficients for firms in the latter sector are smaller, and in some cases insignificant. Firm age is a significant determinant of exporting for firms in the 'computer software development and services' sector, but not in manufacturing. This may be explained by the fact that manufacturing firms are generally older than firms in the 'computer software and services' sector (Berggren et al., 2000), although hypothesis 2a is rejected. Expenditure on R&D is positively related with export intensity for both sectors, and is equally important for both. By contrast, patent income is not a significant determinant of export intensity for either sector, apart from the cloglog model for firms in the manufacturing sector. Finally, receiving advice from the national development agency is positively related with greater export intensity for manufacturing firms, but is insignificant for firms in the 'computer software development and services' sector. This suggests that manufacturing firms may face greater barriers in exporting than services firms, ceteris paribus. ***, ** and * denote significance at the 1%, 5% and 10% levels respectively. Standard errors and p values are reported below the coefficients in round and square brackets respectively. P values are reported for the RESET test.

Discussion and conclusions
Investigating the export intensity of a large sample of independent SMEs, we confirm a number of stylised findings about influential firm characteristics, as well as identifying factors not tested in previous studies. In summary, firm age, size, and R&D Not unexpectedly, firms in the internationally traded sectors are much more export intensive than firms with a greater concentration on the domestic market, such as the retail and wholesale sector. Firm characteristic determinants of export intensity relevant to the total sample are also important for firms in both the manufacturing and 'computer software and services' sectors, although to varying degrees.
The age effect is stronger for firms in the 'computer software development and services' sector than for the whole sample, whilst it is not statistically significant for firms in the manufacturing sectors. This finding is consistent with the behaviour of knowledge based firms that internationalise from an early stage, also known as international new ventures (Knight and Cavusgil, 2005). Our result suggests that, despite a lack of resources typical of young firms, knowledge based firms in the services sector have fewer barriers to exporting and greater ease of access to foreign markets than manufacturing firms (i.e. 'computer software and services' firms generally do not need to establish the physical distribution channels and proceed through the often complex legal and institutional requirements of manufacturing firms).
Thus, consistent with Contractor et al (2007, p.413), "…service firms in general may be able to shorten and mitigate the negative threshold effects of early internationalization far better, and more quickly, than manufacturing companies". This result is also consistent with the finding of Berggren et al. (2000), that manufacturing firms are on average 15 years older than business service firms when evaluated at the median.
The pervasive positive effect of firm size on export intensity is consistent with previous findings that larger firms have greater resources available to invest in export activities (Harris and Li, 2008), and also have greater capacity to absorb sunk costs related to exporting (Ottaviano and Volpe Martincus, 2011). We find the size effect is relatively greater for firms in the 'computer software and services' sector than for manufacturing firms. This result is congruent with the implication that knowledge intensive firms require relatively less investment than capital intensive firms (Love and Mansury, 2009), and can overcome barriers to entering foreign markets more easily (Contractor et al., 2003). Additionally, large firms in the 'computer software development and services' sector operating in countries with a small domestic market need to achieve a high export intensity in order to grow.
We investigate the effect of innovation on export intensity on two levels, considering inputs (R&D expenditure) and outputs (patent income). Consistent with previous studies (Wagner, 2001;Gourlay et al., 2005;Roper et al., 2006), we find a positive relationship between R&D expenditure and export intensity, which supports the technology-based model of export performance (Ganotakis and Love, 2011 our results indicate the importance of investment in R&D and innovation for firms in seeking to gain competitive advantage by developing unique inimitable products and processes. We find that receiving managerial advice and expertise from the national development agency, Enterprise Ireland, has a significant positive effect on export intensity for manufacturing firms and the total sample. Lack of significance for exporting firms in the 'computer software development and services' sector suggests that they rely on internal or alternative external sources. Nevertheless, this finding implies that national development agencies (EPOs in particular) have an important beneficial role to play in supporting exporting firms. This result also suggests that EPOs need to acquire skills and resources in developing and promoting firms developing the latest technologies (e.g. cloud computing), and that they need to be quick to adapt to the needs of SMEs seeking to access new foreign markets. This is particularly true for high-tech firms in small economies seeking to grow. It is important, not only for firms to seek this advice, but for national agencies to inform the independent sector of the services and supports that are available. Our evidence highlights the need for national governments seeking to develop a strong indigenous exporting sector to invest in these services, particularly in light of the lower export propensity of small firms (Roper et al., 2006).
Results suggest that whilst firm-specific characteristics such as size and investment in R&D are equally important, policy makers should provide distinctive supports and services to each sector rather than adopting a uniform approach. Our findings suggest that policy makers can use firm characteristics to identify enterprises that face barriers in internationalisation (Westhead et al, 2002), and thus improve the return on EPO investment by targeting supports more effectively. In contrast with previous studies suggesting segmentation of supports based on owners' experience (e.g. Fischer and Reuber, 2003), we propose a sectoral approach. Appendix 1. Standard errors are in parentheses, p values in square brackets. *, **, *** denote significance at the 10%, 5% and 1% levels respectively.