Standards and export decisions: Firm-level evidence from developing countries

Abstract

Standards and technical regulations set in importing countries have become a rising concern to exporters, especially to those in developing countries. This paper examines the importance of various types of standards in developing-country firms' export decisions. Drawn from the World Bank Technical Barriers to Trade (TBT) Survey database, we find that different types of standards exhibit sharply distinct relations with firms' intensive and extensive margins of exports. Quality standards are positively correlated not only with firms' average export volume across markets and products but also their export scope, measured by the number of export markets and products. A similar relationship is found between labeling requirements and export scope. Certification procedures, however, are associated with a significant decline in the number of export markets and export products. Our results suggest that different approaches should be taken to address each type of technical regulations. Not all standards need to be negotiated away to boost trade, but negotiations on certification procedures with the aim of reaching Mutual Recognition Agreements (MRAs) can help firms improve economies of scale and scope.

Keywords:

• standards
• export decision
• intensive margin
• extensive margin

JEL Classification:

• F13
• F14
• F23

Acknowledgements

The authors are deeply grateful to David Giles, Keith Maskus, Michael Moore, and one anonymous referee for many valuable comments and suggestions. The views expressed here are those of the authors and should not be attributed to the World Bank.

Notes

1. However, as discussed in the second section not all quality standards would have a direct effect on consumer demand. Examples of such standards include environmental standards, which are aimed to reduce consumption (or production) externality (e.g. pollution) and increase the provision of public good. In this case, compliance with standards would increase firms' production costs without affecting their revenue and thus adversely affect firms' incentives to export.

2. These standards consist of standards imposed by National Standards Bodies (NSBs) and de facto standards (i.e. standards set primarily by the market).

3. The details of this dataset are provided in the third section and are also available in Wilson and Otsuki (2003).

4. Maskus and Wilson (2001) provide a comprehensive review of studies in this area.

5. While we consider that all the importing countries of the differentiated product set the same level of the standard, we allow the content of the standard to differ across markets, e.g. the voltage requirement in electrical products. The implication of horizontal differences in national standards for firms' production costs is addressed below. Furthermore, for simplicity, we assume the standard in country 0, the exporting country of the differentiated good, is negligible.

6. Because we allow the content of the standard to vary across importing countries, firms are assumed to pay a fixed cost (e.g. the costs of customizing the product design and performing testing and certification procedures) for each market it exports to.

7. When multinomial-choice variables are inherently ordered, such as voting outcomes, the Multinomial Logit or Probit models would fail to account for the ordinal nature of the dependent variable. The Ordered Probit and Logit models have come into fairly wide uses as a framework for analyzing such responses. We used both Ordered Probit and Ordered Logit models and found no significant difference in our estimation results.

8. The original 15 EU members were considered as one unit in the survey. One disadvantage resulting from this treatment is that the heterogeneity of regulations within the EU cannot be properly explored. A similar concern can arise for the regulatory heterogeneity within a country, say the United States. To deal with these issues we would need more detailed information about firms' export destinations. But it is noteworthy that, in the case of the EU, member countries' heterogeneity in standards and regulations has been significantly decreased by the gradual implementation of harmonization and mutual recognition initiatives.

9. Nevertheless, the majority of firms that were selected after the initial screening interview are exporting firms, which suggest the importance of correcting for potential sample selection bias. We have therefore employed the Heckman (1979) selection model in the fourth section to address this issue.

10. In our sample, the value of the dependent variable ranges from 0 to 10, indicating that firms included in the sample consist of both non-exporters and exporters that export to at most 10 markets. The value of another dependent variable varies from 0 to 15, indicating that the maximum number of products exported by a firm is 15.

11. We also examined the correlations between these standards variables and found them reasonably low. The highest correlation is between testing and certification and has a value of 0.4.

12. A potential concern that may arise in this analysis is the causality between firms' response to these questions and their export status. In other words, it can be the case that only exporters are affected by the existence of foreign standards. However, the data indicates that both exporters and non-exporters have reported the effect of standards on their ability to export, which mitigates the concern of reverse causality.

13. All the above firm attributes, except export history and ownership structure, are measured in natural logs. A firm's export history is a discrete variable that varies from 0 to 4, with 0 indicating no exporting experience and a higher positive number representing a longer exporting history. Ownership structure is also a discrete variable, which is equal to 0 when there is no foreign ownership in a firm, 1 when there is less than 50% foreign ownership, and 2 when the firm is majority foreign owned.

14. Ideally we would also like to include other firm characteristics, such as productivity and R&D intensity, especially in light of the recent growing literature on firm heterogeneity, led by the seminal work of Melitz (2003). However, because only a very few respondents reported information that is required to measure these variables, we are not able to include them in our estimation.

15. This value was obtained by summing up the marginal effects, i.e. dy/dx, for all the respective outcomes (in this case, ). As a benchmark, the average probability of exporting, as reported in the second column of Table 4, is 0.98 .