ICT Use and Exports. Patricia Kotnik, Eva Hagsten. This is a working draft. Please do not cite or quote without permission of the authors.

Size: px

Start display at page:

Download "ICT Use and Exports. Patricia Kotnik, Eva Hagsten. This is a working draft. Please do not cite or quote without permission of the authors."

May Norman
5 years ago
Views:

1 ICT Use and Exports Patricia Kotnik, Eva Hagsten This is a working draft. Please do not cite or quote without permission of the authors. September 2012

2 Introduction Studies have shown that two major distinguishing characteristics of high-growth SMEs are their export orientation and their innovative capabilities (OECD, 2010) and the role of export in firm performance is not limited to SMEs only (see, for example, Wagner, 2007). Given the current economic situation in Europe, the issue of enablers helping in opening up global markets to firms is increasingly relevant. The role of ICT in this process remains relatively unexplored in the literature. Research on economic impact of ICT related to globalization has so far mainly focused on multinationals and the role of trade in technology adoption. The role of ICT in exporting of firms remains relatively unexplored. Some evidence is starting to build up, particularly in international business and international entrepreneurship literature, showing ICT as relevant for internationalization of firms, especially smaller ones. Existing research has predominantly used relatively small samples, used qualitative research or focus on groups of firms with specific characteristics. A study within the ESSLimit project, based on comparable data across a number of countries and using larger samples and common methodology, is thus a unique opportunity to contribute to this topic. Our study will add to this area of research by empirically investigating factors that influence export behaviour of firms. The available data allows us to link datasets from ICT use surveys, Community innovation surveys, and structural business statistics and to estimate the model on firm-level data for 12 countries. The main contribution is thus to address the role of ICT use in exporting in a cross-country perspective using common methodology and comparable datasets. Previous findings When positioning themselves on international markets, advantages to ICT use can be manifold (apart from the impact ICT use has on improvements in productivity and innovation performance which are also relevant for firm exports). One is a possibility of a direct link with customers and a reduced need to establish a physical presence abroad which results in a decrease in transaction costs. A number of marketing tasks can also be fulfilled through ICT (like international advertising, communication with customers, order management). Strategic use of internet can thus be a form of foreign market entry for smaller firms by which they compensate for disadvantages connected to weaker physical presence on foreign markets when compared to large firms and by which they access parallel foreign markets more quickly. A study on the speed of internationalization of SMEs has established a strong relationship between ICT use and rapid international growth (Morgan-Thomas and Jones 2009). A growing body of work studying born-global firms also shows that Internet helps them to grow sales rapidly and reduces liability of newness and foreignness and resource scarcity (Reuber and Fischer 2011). (will be added) The data and estimation methods Export data have been gathered for 12 countries taking part in ESSLimit project, using various sources available to national statistical offices. Data on value of exports were used and based on this two variables were derived: i) export dummy (indicating that firm sells its goods and services outside its own national boundaries); and ii) export intensity (defined as a ratio between value of export and nominal gross output). Not all data is available for all of the countries included in the analysis (Austria, Italy, Netherlands and Norway data include export of goods only whereas in other countries data are available for exports of goods and services). Through examining the results, some inconsistencies in the data across countries were detected

3 so for the sake of comparability the data for some countries had to be left out (export dummy in the Netherlands and export intensities in Ireland and Italy). Source of ICT use data is the E-commerce survey, an annual business survey measuring the use of ICT by firms. To capture ICT use that might be connected to internationalization we use the following variables: online presence (dummy for firm having a website), online transactions (dummy for E-sales firm selling through website or Electronic Data Interchange), and share of workers with access to fast internet capacity (proportion of broadband Internet enabled employees). To undertake analysis that includes other determinants of exporting, other sources of data were added. A set of production variables (employment, labour productivity, capital/labour ratio, wages, age, dummy for foreign ownership) were sourced or derived from Business Register and/or Structural Business Survey. Availability of human capital data varies between countries and to make full use of the data different variables (and thus sources) were used between countries. In some countries, human capital is represented by percentage of workers with post upper secondary IT education (HKITPCT) and by percentage of workers with post upper secondary non-it education (HKNITPCT); in one country it was captured by percentage of workers with post upper secondary education (HKPCT); and in other countries wages were used as a proxy for human capital. Dummies for product and process innovations are included to measure innovation activities of firms, for which Community Innovation Survey was used. The merging of the different datasets results in a relatively small sample for each year therefore we have used independently pooled data across the period that was available for each country. For most countries this period refers to (the last available Community Innovation Survey data refers to year 2008). The dataset that is built using the production variables as well as e-commerce and innovation surveys (the so-called PSECIS sample) excludes many firms for which all variables except for innovation are actually available. Also, it is biased towards larger firms (due to the sampling used in innovation surveys). Therefore, a dataset that excludes innovation data (the so-called PSEC sample that merges production variables and e-commerce data) is also used for analysis. This dataset covers the period for most countries. However, the estimates based on PSEC sample need to be interpreted with caution: not including innovation variables in the model leads to an omitted variable problem (and thus biased and inconsistent estimates) since we can expect that ICT use and innovation variables are correlated (see for example van Leeuwen and Farooqui, 2008). The analysis includes two steps: in the first, the probability of entry into exporting is studied, whereas in the second step export intensity and export intensity growth are estimated. To answer the question whether ICT use has an impact on whether a firm exports or not, we have estimated a probit model. To this, we have added OLS estimates of export intensity and export growth, applied to a full sample (including exporters and non-exporters). 1 These sets of estimates were applied to the PSECIS sample and also to PSEC sample. In all regressions, controls for industry and year effects were included. Endogeneity issues need to be addressed since differences between exporters and nonexporters cannot tell us whether good firms become exporters or exporters become good firms. Given the nature of the data available we use lagged values of explanatory variables to mitigate this problem. 1 For three countries, the dataset on export intensity included missing values for non-exporters and not value 0. The resulting datasets included exporters only, thus allowing for the possibility of selectivity bias. This was addressed using the Heckman method. The results for Austria, Norway and Sweden, reported in Table 2, thus refer to the results of outcome equation of the two-step Heckman model. (One of the recommendations of ESSLimit project as a whole is for Eurostat to develop a standard for treating missing and zero values in surveys across Europe.)

ICT use and exports of firms: Indications from descriptive statistics Basic comparisons between exporters and non-exporters in our dataset show that the following are higher for exporters: labour

4 ICT use and exports of firms: Indications from descriptive statistics Basic comparisons between exporters and non-exporters in our dataset show that the following are higher for exporters: labour productivity, share of highly skilled human capital, and the likelihood to be innovative (as measured by the introduction of product or process innovations). These conclusions hold for all groups of industries and countries (with a few exceptions). Figure 1 shows labour productivity of exporters and non-exporters in year 2009 for the countries included in the analysis (and is based on the PSEC sample as are other figures in this section, except those including innovation data which are based on merged production and innovation survey samples). 2 Descriptive statistics are shown for groups of industries since this allows us to gain additional insights concerning characteristics of firms and their ICT use. A project specific industry aggregation is used which has an advantage of splitting ICT producing industries from ICT using industries. Data is thus split across three groups of industries, the first grouping the ICT producing ones: electrical machinery, post and communication services (Elecom); manufacturing sector excluding electrical machinery (MexElec); and market services, excluding post and telecommunications (MServ). The comparison shows that ICT producing industries are more productive than the other two industry groups. Also, productivity difference between exporters and non-exporters also seems to be largest. Figure 1. Labour productivity of exporters and non-exporters, 2009 Source: ESSLimit cross country dataset In Figure 2, human capital of exporters and non-exporters is shown, measured by percentage of workers with post upper secondary education. Exporting firms have higher human capital in all countries and industries (except for ICT producing industries in France) but the differences to 2 The countries denoted with faded colours (on the right side of the figures) have data available on exports of goods only which is not directly comparable to other countries for which data on exports of goods and services are used. This holds for all figures in this section.

non-exporters are smaller than observed for labour

The share of highly educated employees is the lowest in

non-exporters, 2009 Source: ESSLimit cross country dataset

amongst exporters and nonexporters shows that innovators

5 non-exporters are smaller than observed for labour productivity. The share of highly educated employees is the lowest in manufacturing sector excluding electrical machinery. Figure 2. Proportion of highly skilled human capital of exporters and non-exporters, 2009 Source: ESSLimit cross country dataset Comparison of share of firms engaged in product innovations amongst exporters and nonexporters shows that innovators are much more frequent amongst exporters, especially so in ICT-producing industries (Figure 3). Similar conclusions can be drawn for firms with process innovations. Figure 3. Proportion of firms with product innovation, exporters and non-exporters, 2008 Source: ESSLimit cross country dataset

When comparing ICT use of exporters and non-exporters, descriptive statistics show that the former use ICT more intensively; however, this

Whereas this share is higher for exporters in all countries for manufacturing (excl. electrical machinery) and services (excl.

exporters and non-exporters that engage in e-sales is shown in Figure 5.

6 When comparing ICT use of exporters and non-exporters, descriptive statistics show that the former use ICT more intensively; however, this cannot be generalized to all industries. Figure 4 shows the comparison of proportion of broadband Internet enabled employees in Whereas this share is higher for exporters in all countries for manufacturing (excl. electrical machinery) and services (excl. post and communication) this does not always hold for ICTproducing industries (which could be connected to the saturation when it comes to this ICT use variable). Figure 4. Proportion of broadband Internet enabled employees of exporters and non-exporters, 2009 Source: ESSLimit cross country dataset Share of exporters and non-exporters that engage in e-sales is shown in Figure 5. The differences between these two groups are the largest in non-ict producing industries (Mexelec and Mserv). In the case of ICT-producing industries the differences are less pronounced or even show the situation where non-exporters engage in e-sales more frequently. Figure 5. Proportion of of exporters and non-exporters with e-sales, 2009

7 Source: ESSLimit cross country dataset Lastly, we show the average export intensity for the three groups of industries (Figure 6). Manufacturing sector excluding electrical machinery sells the highest share of output on the foreign markets but ICT producing sector is not far behind, at least in some countries. Not unexpectedly, market services have the lowest export intensity, but export is still important for this sector in small open economies like Slovenia, Luxembourg, Ireland and Denmark. In the case of Ireland the export intensity of manufacturing sector is extremely high (implying the share of exports in gross output is higher than 100 percent) which is apparently explained by merchanting. The differences in export intensities between countries reflect the macro level factors that influence the countries openness in trade. Figure 6. Export intensities in 2009 Source: ESSLimit cross country dataset On the basis of descriptive data we can conclude that exporting firms on average use ICT more intensively, as measured by broadband Internet enabled employees, e-sales and online presence. However, we cannot conclude that there is a relationship between ICT use and exports of firms since exporters are characterized by higher firm performance therefore their ICT use is also higher. We need to control for firm characteristics other than ICT use. The results of regression analysis are presented in the next section.

8 The main findings The main results of the estimates are shown in Tables 1 and 2. The overall conclusion can be drawn that ICT use does contribute to exporting behaviour of firms in some countries, but not in all of them. The results seem to imply that ICT use helps the firms in entering the foreign markets, whereas the evidence of its contribution to export intensity is mixed. Let us first comment on the probit estimates. The results (shown in Table 1) show that past export behaviour (capturing sunk costs connected to entering foreign markets), size and human capital, as well as product innovations increase the probability of entry into exporting in most countries. Online presence of a firm (having a website) increases the probability of entry into exporting in Ireland, Italy, Norway and Slovenia. E-sales variable is statistically significant (with a positive sign) in Finland, France, Sweden and Slovenia; and broadband Internet enabled employees increase the probability of exporting in Norway and UK. These results are confirmed when the PSEC sample is used for estimates. Focusing on ICT use variables, they are now statistically significant in more countries (in all countries except Denmark and UK for online presence; in Italy and Sweden as well as in Norway and UK for broadband Internet enabled employees). However, as pointed out before, the impact of ICT use variables might be overestimated since innovation is not included in the model. In the case of e-sales, it now seems it increases the probability of exporting in the case of Finland, France, Italy and UK. Turning next to the results of the OLS estimates, in Table 2 we report on results of the model of export intensity (using the PSECIS sample). Results show that in most countries there is a positive and significant relationship between human capital and also labour productivity and export intensity. Focusing on ICT use variables, we could conclude that firms with lower ICT use have higher export intensity (online presence is statistically significant in Norway, e-sales in Finland, Netherlands, and Norway, and broadband connectivity in Austria and Norway; in all cases the sign is negative except in UK where broadband connectivity is positive). When interpreting these results, one needs to keep in mind that PSECIS sample is biased towards larger firms and it is possible that these firms focus more on B2B market where other laws apply (online presence and e-sales are probably not really relevant for their export intensity). Looking at the results for PSEC sample the picture is somewhat different: online presence is statistically significant in France (positive) and Norway (negative), e-sales in Finland, Norway and UK (negative), and Luxembourg and Sweden (positive), and broadband connectivity in France, Luxembourg, Sweden and UK (positive) and Austria (negative). Finally, the results of the model examining growth in export intensity indicate that in France and Luxembourg the firms with higher broadband connectivity have higher export growth.

9 Table 1: Estimates of probit model (PSECIS sample, Total economy) dependent variable: exporting undertaken or not AT DK FI FR IE IT LU NO SE SI UK lagged exports + *** + *** + *** + *** + *** + *** + *** + *** + *** + *** + *** log labour productivity + ** - *** log employment + *** + *** - *** + *** + *** + ** K/L + *** HKITpct + *** HKNITpct + *** + *** + *** HKpct log wages + *** Age product innovation + ** + ** + ** + ** + *** process innovation firm has website + *** + *** + *** + *** firm sells through website or EDI + ** + *** + ** + ** broadband % + *** + *** foreign ownership + *** + *** + ** + *** Intercept + **

10 Table 2: Determinants of exporting intensity (PSECIS, Total economy) dependent variable: export intensity AT DK FI FR LU NL NO SE SI UK log labour productivity - * + *** + *** + *** + ** + *** log employment + *** - *** + *** - ** HKITpct + *** + *** + *** + *** + *** HKNITpct + *** + ** + *** + *** HKpct log wages Age - ** + ** + *** + *** product innovation + *** + *** + ** + ** process innovation + *** + ** firm has website - * - *** firm sells through website or EDI + * - *** - ** - ** broadband % - ** - ** + *** foreign ownership + ** + *** + *** + *** + *** Intercept + * - *** - *** degrees of freedom R

11 Conclusions This study has tried to assess the role of ICT as an enabler of internationalization of firms. We have used comparable firm-level data from E-commerce survey, Community Innovation survey and a set of production variables for 12 countries to study determinants of export performance. Common methodology was used to estimate a model of the determinants of firm entry into exporting and of proportion of output that is sold outside of domestic market. To capture ICT use that might be connected to internationalization we have used the three variables: online presence (firm having a website), E-sales, and share of workers with access to fast internet capacity. Our results suggest that ICT use does contribute to exporting behaviour of firms in some countries, but not all of them. In a larger number of countries firms with higher ICT use (especially when measured by online presence and also e-sales) are found to have higher probability of exporting so the results seem to imply that ICT use helps the firms in entering the foreign markets. The evidence of its contribution to export intensity is mixed, however. Each of the three ICT use variables is statistically significant in a few countries, in some indicating a negative and in other countries positive relationship between ICT and export intensity. In interpreting the results, one should bear in mind the nature of the research setup which is still exploratory. Suggestions for further research include a stronger focus on the mechanisms through which ICT use affects exports which would require taking into account the contribution of ICT use to productivity and innovation output (and thus the use of structural equation modelling). Another interesting venue for research would be to explore in greater detail the differences between the industries, specifically the differences between industries focusing on B2B and B2C markets and also the specifics on ICT-producing industries. And, last but not least, the reasons for the differences between countries would be interesting for further research; that is, why does ICT use contribute to exports of firms in some countries but no others.

Flexible Working Arrangements, Collaboration, ICT and Innovation

Flexible Working Arrangements, Collaboration, ICT and Innovation A Panel Data Analysis Cristian Rotaru and Franklin Soriano Analytical Services Unit Economic Measurement Group (EMG) Workshop, Sydney 28-29