Conclusion

---

The objective of this study was to identify the key factors that influence adoption of EDI in the specific context of SME. Based on previous studies in the related literature including adoption and diffusion of innovation and related EDI and SME studies, a research model was developed. Seven factors, namely, perceived direct benefit, perceived indirect benefit, perceived cost, IT knowledge, management attitude towards EDI, government incentives and enforcement and trading partners' influence, were posited as important in the adoption decision. Seven hypotheses were proposed to examine the relationships of the factors with EDI adoption. Based on responses from 644 SME, five factors were found to be important in distinguishing early EDI adopters from late adopters. The five factors are perceived direct benefits, perceived cost, IT knowledge, government incentives and enforcement and trading partners' influence.

One of the main contributions of this study is that it is one of very few research studies that examined EDI adoption in an SME environment using the survey approach, which, at least theoretically speaking, can increase the generalizability of findings in previous studies that mostly used the case study approach. Compared to Raymond and Bergeron's (1996) study with only 39 SME, this large-scale empirical study can provide more accurate assessments and implications for factors affecting EDI adoption in SME that should be of great importance to an SME's strategic decision concerning the adoption of EDI. The results are in general consistent with findings in the innovation adoption literature, as well as related EDI and SME studies, suggesting that findings from previous innovation adoption studies for large organizations might also be applicable to SME.

As with all empirical studies, there are several limitations that, however, must be taken into consideration. First, the study was "ex post facto" research, which always has a methodological problem of requesting respondents to look back in time to explore possible causes and relationships of the effects on the dependent variables such as an adoption decision. To overcome or to minimize the seriousness of this problem, in our study, the respondents were reminded that contextual factors were to be measured prior to the adoption decision. Note that the early adopters were asked to provide the ratings/ measurements at the time of adoption, whereas late adopters were asked to provide the feedback as of the time of the study. However, we cannot be completely sure that the respondents' answers were not influenced by the experience of EDI adoption. Also, since the early adopters were asked to provide the ratings at the time of adoption, whereas late adopters were asked to provide the responses as of the time of the study, this time difference might be a cause or reason why trading partners' influence is negatively correlated with EDI adoption. As argued above, when the early adopters decided to adopt the technology, there were few within the industrial community that were using it. Trading partners' influence, if any, should be little. Contrarily, late adopters might receive more influence from their trading partners as many might have already adopted the system in this late stage.

Second, all the data in this study were obtained via self-reports from a single respondent in each SME. The data thus possess the potential bias for inflated correlation due to common method variance. Applied multi-item indicators for each contextual factor in our study were therefore used to lessen the common method variances. Also, results of validity and reliability tests help to provide sufficient confidence levels for each contextual factor in this study. The single respondent, key informant method might also be a problem. However, since in general, the size of SME is relatively small and the structure of SME is rather simple, the owner, who in most cases was the key informant, could be comfortably assumed to be knowledgeable of the EDI adoption decision process and of related organizational information and his/her responses would be representative of the factors.

Third, the EDI service examined in this study was a business-to-government type of EDI. Findings of prior studies on the role of power and trust between firms on the adoption and use of EDI (for example, Hart and Saunders, 1997) may therefore be difficult to compare with our findings, and vice versa. In other words, interpretation of the findings should be made with care when extending them to those EDI services/products of a business-to-business nature. Also, the business-to-government nature of the EDI service being studied here precluded us from looking at the influence of competitors on EDI adoption in small businesses.

The study was conducted in late 1998 when Internet-based EDI systems were still not popularly advocated by major EDI industry players. Some of the significant factors found in this study, such as perceived costs, might become less of an issue when small businesses begin to conduct EDI via the Internet. Other significant factors, such as perceived direct benefits and government incentives and enforcement, however, are believed not to be impacted by the change from value-added networks to the Internet as these factors focus on the values/services provided by the technology, rather than the "operating platform". IT knowledge might become more of an issue as small businesses now have to cope with another operating platform, the Internet, with which they might be even more unfamiliar. On the other hand, this factor might also become less of an issue as Internet-based platform is more universal and hence businesses are more likely to find knowledgeable IT staff. It should be useful and interesting to test the validity of the research model again with an Internet-based EDI system.

Finally, the results in this study provide a useful understanding of EDI adoption in the SME context. As this study was conducted on SME in Hong Kong, caution should be taken when generalizing the results to SME in other countries. Different cultural contexts and institutional policies may cause differences or variations in outcomes. The above "culture-related" factors may or may not be related to an organization facing the adoption decision. Thus, replication in different countries in the context of SME would greatly improve the external validity of the findings.

Implications for Research

Despite the above limitations, the findings of this study have several research implications. First, the study examined several variables that might influence EDI adoption. The factors and relationships outlined provide the foundation for the development of a more robust and rigorous model for future research. Rigorous validation of the contextual factors in replicating this study is suggested. Also, one limitation of a parsimonious model is that some additional significant factors are not included in the model. Other factors identified by other related studies may be useful to increase the explanatory power of this research model and to provide a more extensive or complete model. For example, Thong (1999) suggested that external IT expertise such as vendor support or consultants helped to explain adoption and implementation of IT in SME. Besides external expertise, factors like CEO involvement, organizational structure and organizational culture, which are possible influences on EDI adoption in SME, are suggested to be included in future research.

The study found that governments could play vital roles in the adoption and diffusion of new technologies. This study, however, did not assess how and exactly what a government can do to influence the adoption and diffusion process. Government influence may vary from country to country and different policies may help to promote adoption of EDI to different extents. Therefore, governments should choose policies that fit the specific institutional and industrial environment. To what extent the government should be involved in intervention and actions in order to maximize adoption of EDI or new technologies is an interesting and important issue. Thus, a detailed investigation of government influence in the adoption and diffusion of an innovation can provide insights into the promotions of innovations.

A related issue is the relationship between dependency on government/trading partners and EDI adoption. Our study looked at an EDI-focused on trade declarations, which are required for firms doing import/export trading business. The partnership between the firms and the government is mandatory. In other business-to-business or non-mandatory business-to-government types of EDI, this "dependency" factor would be an important construct to be included in the analysis.

The finding of a negative relationship between the importance of trading partners' influence and small business EDI adoption is both interesting and thought provoking. The rationale we have provided for this "unexpected" result is that since the adoption and/or diffusion of EDI in the small business sector was still at the early stage with very few firms using it, there was just not enough critical mass to make this factor influential. This suggests an interesting and possibly fruitful line of research that investigates the validity of the theoretical model and the significance of the factors involved along the "life cycle" of the diffusion of the technology. A longitudinal study that examines the model at different stages of the diffusion of a technology is recommended.

EDI is a kind of technology that is sitting between "old" technologies using ANSI X12 or EDIFACT standards and "new" Internet-based XML or Open EDI platforms. It has been estimated that while VANs carried about 95% of all EDI volume in 1997, half of the volume might be moved to Internet-based platforms by 2002 (Wilson, 2000). It therefore would be interesting to investigate how firms, both large and small, move from using an old technology to a new technology which basically provides the same services, from a technology adoption perspective.

Organizational innovation is explained through a stage model. This study examines only the adoption stage. Other interesting research issues would be to examine the impact of these factors on the next stage of the adoption-diffusion process, the impact of the technology (Iacovou et al., 1995). Many researchers have argued that different impacts or effects of factors exist in different stages of the adoption-diffusion process (Cooper & Zmud, 1990). For example, while the investment cost of adopting an innovation is a factor that has a negative influence on adoption, there is evidence that it can have a positive influence in the implementation stage. An organization is unlikely to adopt an innovation that requires high investment. Once it decides to adopt, high investment cost will act as a motivator for an organization to diffuse the innovation further inside the organization because more benefits may be obtained from wide diffusion of the innovation and these finally may be commensurate with the investment cost. It would be useful to determine whether the same set of factors that are important in adoption are relevant as determinants in the diffusion stage.

From a related but slightly different perspective, the focus of the study was on adoption of EDI, which should be considered as part of a broader issue, the management of technology. As adopting a technology should eventually lead to some form of positive impacts or advantages, which can be operational, managerial, and strategic in nature, for an organization, evaluation and assessment of the adoption and implementation of the technology are important too. Added work should therefore be done to extend the model examined in this study to include integration and impact, as suggested by Iacovou et al. (1995).

Implications for Hong Kong SME

Because of the different profiles of industries in Western countries and Asian countries, EDI adoptions in Asian countries have been relatively slower than in North America and Europe (Burn, 1995). Within Asia, EDI adoption in Hong Kong further lags behind its competitors. This study shows that the government's intervention is crucial in EDI adoption. To foster the adoption, governments may direct their own agencies to adopt EDI and require organizations to communicate with them through the system in order to promote the use of EDI. In the context of Hong Kong, Tradelink, with the Hong Kong Government as the largest shareholder, is the largest EDI services provider for providing electronic communication for trade-related documents between the Government and Hong Kong organizations. The above finding reinforces the decision of the Hong Kong government to require all organizations in Hong Kong to use electronic submission of trade declarations through Tradelink by mid-2000. Under direction from the government, this mobilization of adopting EDI helps SME in Hong Kong to reach the "critical mass", thus overcoming the network externality problem of EDI adoption.

In addition to being a mandatory service, the success of TradeNet in Singapore was also attributed to a range of government subsides and technical backup provided to support the EDI adoption (Teo et al., 1997). Iacovou et al.'s (1995) EDI study also found evidence of subsidies in the early stage of expansion facilitating faster adoption by SME. The Hong Kong government has also been providing various kinds of financial subsides and technical supports to SME through its Industry Department, Trade Development Council, Productivity Council, etc. SME should make full use of this assistance in order to overcome the substantial burden, both financial and technical, especially in the early stage of the adoption.

Moreover, the finding of the importance of perceived direct benefits on the EDI adoption decision also suggests that EDI initiators or service providers in Hong Kong, including Tradelink and other relevant organizations and institutions, should make greater efforts in marketing EDI technologies to let the SME appreciate the benefits of adopting EDI. This kind of greater exposure to the potential benefits and impacts of EDI on business operations and greater knowledge about the technology may lead to faster EDI adoption by SME.

In conclusion, it is believed that a better knowledge of executive decisions about adoption of information technology in small businesses should be helpful to their practicing managers in understanding the qualifying factors for which EDI is most appropriate to their organizations. This study has obtained some interesting findings that advance our understanding on this issue.

---