Trading ain’t easy: How complex is it to trade goods?

Carlo Altomonte, Gábor Békés

19 November 2010

a

A

A large economic literature as well as a number of policy success stories point to the superior performance of exporting firms in terms of size, wages and productivity, and thus the importance of an adequate openness to trade flows as a driver of competitiveness (see for example Yang 2009). More recently, evidence has been found that importers also have a variety of positive attributes, being bigger and more productive than non-importers, with both imports and exports appearing to be highly correlated and concentrated among firms. In addition, it has also been shown that importing intermediate goods tends to improve plant productivity in a number of countries.

All the above evidence is consistent with the increasing role acquired by global value chains and international production networks in the world economy. And yet, notwithstanding these findings, national trade policies to date still revolve around export promotion programmes, while they tend to neglect the relation between importing and exporting activities, the nature of traded products and the ensuing implications for firms’ performance.

New evidence on the benefits of exports – and imports

In a recent paper (Altomonte and Békés, 2010), we show that in a small open economy increasingly integrated into global value chains, Hungary, both importing and exporting activities are closely related to productivity, suggesting that more productive firms self select into both exporting and importing. Thus, ignoring the import activity of an exporting firm might actually lead to a biased estimation of the productivity premium attributed to exporters, hence an excessive emphasis given to export promotion policies compared with policies aimed at integrating domestic firms into international networks of production.

Another issue for policy is the composition of goods a firm imports and/or exports. This stems from our finding that the productivity of trading firms is heterogeneously related to various dimensions of “trade complexity”, that is, different characteristics of the bundle of goods that firms import or export. Complexity may be defined by factors such as the number of products and countries involved in the firms’ international network, the quality and contractibility of the traded goods or the degree of substitutability with local products. These factors turn out to all be correlated with the productivity of trading firms – to a varied extent.

Imports, exports and productivity premia

To perform our analysis we exploited the availability of detailed product-country international transactions data of Hungarian manufacturing firms observed yearly from 1992 to 2003 (for a total of some 192,000 firm-level observations). As our data refer to import and export transactions which can be directly matched to the operations of manufacturing firms, we excluded transactions which take place through the activity of an intermediary or a wholesaler. Hence, we were able to directly relate the characteristics of each trade transaction to the firm’s productivity.

First of all, we find evidence of a self-selection effect of the most productive firms into both the importing and exporting activities. In Figure 1 we show the average total factor productivity (TFP) premium of firms categorised by trade status with respect to the average TFP of the entire sample. As can be seen, the importing status of a firm, on top of its exporting activity, clearly partitions the sample in three sub-groups, ranked in terms of TFP. This ranges from no importer (whose productivity is 21% below the average firm in the sample) to new importer (14% more productive) to permanent importer (42%).

Figure 1. TFP ranking by import status (difference from sample mean TFP)

Note: Simple average of total factor productivity calculated at the firm level. Ranking is robust to the choice of methodology.

More in general, when taking into account the importing status of exporting firms in a multivariate regression, controlling for a number of time-varying firm-level characteristics as well as industry and time fixed-effects, the productivity premium of exporting firms traditionally found by the literature remains positive and significant, but it is greatly reduced, from around 36% to 15% in our baseline specification. In other words, failing to control for the correlation between importing and exporting activities within a firm, as a large part of the debate has done so far, might lead to a significant upward bias in the estimated productivity premium of traders.

Trade complexity and productivity

In order to explore the heterogeneity in productivity premia and its potential drivers, we derived from our transaction data a set of proxies to technological and relationship-specific dimensions of the trading activity, which we generally refer to as “trade complexity”. We have defined trade complexity as a combination of three distinguishing features: quantity, quality of products and technology:

  • the quantitative dimension of complexity is related to the operational and organisational costs incurred by a firm when undertaking a trade activity, and it is proxied by the number of countries and products a firm has to deal with (Eaton et al. 2004);
  • the qualitative dimension of complexity is related to the costs associated with writing contracts for specific products which need to be screened for quality and insured against the risk of a faulty delivery (for the importer), or the monetary risk of not being paid (for the exporter) (Berkowitz et al. 2006). We found that the qualitative features of complexity may well be proxied by the average unit value of the traded bundle (UVT).
  • the technological dimension of complexity deals with the imperfect substitutability of domestic inputs compared with foreign ones (Halpern et al. 2009) and the associated search costs for the “right” input available on international markets. Our key proxy is called a “substitutability index”, measuring the average number of countries from which a specific product is imported from / exported to. Firms exporting very specific, low-substitutability goods might require particular production processes or specialised channels for their sale and hence, a lower value is associated to a higher trade complexity.

In addition to these three different dimensions of complexity, we also control for other possible drivers of trade costs, such as the (weighted) average distance from which the bundle of products is traded as a proxy for transport costs and the (weighted) average size of the partner countries, as measured by a country's population, as a proxy for marketing/search costs (Arkolakis 2008).

Table 1 displays the descriptive statistics for our main indicators of complexity measured at the firm level, across different categories of trading firms. The first two columns look at all the firms’ export and import bundles. The importing side of trade seems to be associated on average with more complex bundles, along all the three measured dimensions of complexity. We find that the average imported bundle contains three times more products, with twice the average unit value, and is traded with 20% more countries, than the average exported one. Moreover, the substitutability on individual products is lower for the imported bundle. Imports also come, on average, from more distant (by 31%) and larger (35%) countries with respect to exports.

To confirm this, we considered firms that started to export or imported (switchers – columns 3 and 4) as well as compared the exported and imported bundle of firms that trade both ways (columns 5 and 6). Differences remain rather strong.

Table 1. Complexity indicators: descriptive statistics

We argue that self-selection matters in terms of exporting as well as importing and that the process is related to the complexity of eventual bundles of trade. Thus, we regress ex-ante productivity of trading firms on our indicators of complexity only to find that complexity account for an additional third of the overall productivity premium. In particular, adding complexity measures to our baseline productivity regression, these measures are significant while the productivity premia are reduced from 15% to 10% for exporters (already controlling for the upward bias induced by the import status) and from 45% to 31% for importers. Importantly, we also find the elasticity of productivity to a change in our trading complexity indicators to be larger for importers than for exporters, and to be different for different indicators of complexity. Results also hold when controlling for switching firms or for different proxies of complexity.

Conclusions

Our results suggest that various dimensions of “trade complexity”, that is different characteristics in terms of quantity, quality and technology of the bundles that firms import or export, correlate differently with productivity across firms’ trading activities (importing, exporting or both). Due to this heterogeneity, policy actions should thus address the interaction between the diverse facets of trade complexity and the importing/exporting activities of the firm, rather than a firm’s generic trade status.

For example, trade participation can be boosted via educating people with the skills that allow handling complex processes, such as languages, communication, engineering, legal services, and management.

Such a more specific policy focus is likely to be more effective in stimulating firms’ productivity, eventually fostering their better integration into global value chains.

References

Altomonte, C and G Békés (2010), “Trade Complexity and Productivity” CeFIG Working Paper No. 12, October 2010.

Arkolakis, C (2008), “Market Penetration Costs and the New Consumers Margin in International Trade”, NBER Working Paper 14214.

Berkowitz, D, J Moenius and K Pistor (2006), “Trade, Law, and Product Complexity”, The Review of Economics and Statistics, 88:363-373.

Bernard, A, JB Jensen, S Redding and PK Schott (2010), “Intra-Firm Trade and Product Contractibility”, American Economic Review, Papers and Proceedings, 100:444-448.

Eaton, J, S Kortum, and F Kramarz (2004), “Dissecting Trade: Firms, Industries, and Export Destinations”, American Economic Review, Papers and Proceedings, 94:150-154.

Halpern, L, M Koren, and A Szeidl (2009), “Imports and productivity”, CeFIG Working Paper No. 8, Budapest.

Yang, Dean (2009), “Does exporting improve firm performance?”, VoxEU.org, 24 March.

a

A

Topics:  International trade Productivity and Innovation

Tags:  globalisation, Export, import

Associate Professor of Economics, Bocconi University and Visiting Fellow, Bruegel

Research Fellow, Institute of Economics of the Hungarian Academy of Sciences, Budapest

Events