Shedding some light on dark matter: Trade costs in services

James Anderson, Ingo Borchert, Aaditya Mattoo, Yoto Yotov

21 October 2015



Almost half of value added in exports is contributed by services, and developing countries have steadily increased their participation in trade in services to reach 38% in 2013 (United Nations 2015).  Given the economic importance and increasing significance of services trade, surprisingly little is known about trade costs and border barriers in services (Francois and Hoekman 2010).  Policy barriers to services trade comprise relatively opaque and hard-to-measure regulations, as for example in professional and financial services, rather than transparent and numerical measures like tariffs.  Transport costs in services are nebulous, reflecting costs such as those of electronically delivering business services, in contrast to freight rates on tonne-mile shipments of physical goods.  In the absence of explicit and quantitative measures of either protection or transport costs, an analytical model of trade flows can help reveal information about barriers to services trade.

In new research (Anderson et al. 2015), we apply recent developments in the gravity literature to a new multi-dimensional data set of services border barriers at an unprecedented level of detail (12 service sectors and 28 countries over the period 2000-2007).  The sectoral and geographical coverage as well as the inclusion of intra-national trade flows – crucial for estimating border effects – sets our dataset apart from those used in previous gravity estimations and enables us to shed new light on the murky effects of globalisation on services trade.  The resulting border effect estimates for 28 countries and eight years are probed for determinants in a second stage cross-country panel regression.

Borders in services trade: Large and heterogeneous

We find that border barriers in services trade are large, significant, and vary widely across countries and sectors.1 The trade volume effects of borders are identified as gravity regression estimates of the coefficients (in logs) of indicator variables that take a value of one for intra-national trade.  Strictly speaking, the border effect is relative, i.e. the effect on internal trade of border cost relative to internal cost and for each country relative to a global average of such effects.  A high border cost relative to internal cost raises internal trade.  International trade costs are estimated as usual from international flows controlling for all standard gravity covariates (e.g. distance, common official language, etc.).

Across countries, border barriers decline with economic size (Figure 1).  An estimate of one means that internal trade is 2.72 (equal to e, the basis of natural logarithms) times larger than it would be with a zero border estimate, all else equal.  The largest estimates are for small countries like Slovakia, Estonia, Lithuania, Latvia and Slovenia, and the smallest are for Great Britain, Holland, Canada, Germany and Austria.  The inverse relationship between economic size and trade costs in services trade is consistent with, and complements, the findings of Waugh (2010), who shows that less developed countries face higher aggregate trade costs. (The US is the only country that exhibits overall negative border effects, mostly driven by the travel/tourism sector. Rounding the US border estimate down to -2, the implication is that internal trade is 0.135 times the value it would have if the US border estimate were zero.)2

Figure 1. Country size and border effects in services trade

The largest border effects are observed in the financial, insurance, and research and development services sectors (Figure 2).  The high estimates for financial and insurance services reflect the fact that these services are not highly tradable across borders (Jensen and Kletzer 2005) and tend to be produced and consumed domestically.  It is interesting to see that the business services sector (BUSIN) exhibits relatively low barriers to cross-border trade, reflecting the influence that ‘business process outsourcing’ and related developments have had to lower revealed border effects.  Borders matter least for transport, travel and communications services, as one would expect, since these services inherently involve substantial transactions across borders.

Figure 2. Sectoral border effects in services trade

Uneven globalisation in services trade

Even over the limited period of time covered here (2000 to 2006), border barriers in services trade have fallen, reflecting the increasing globalisation of services.  But the decline varies considerably across sectors and across countries – border barriers have fallen in all sectors, but more so in sectors with lower initial barriers.  The negative correlation between the initial level of border barriers and the decrease is close to 0.9.  Thus, border barrier heterogeneity across sectors has actually increased.  Intuitively, those sectors with initially low border effects – transport, travel and communications – are precisely the ones that have been most strongly affected by improved communications and technology. In contrast, sectors like finance, insurance and audiovisual services may also have seen a positive technological impact, but are hampered by serious regulatory impediments. 

Perhaps more importantly, the change in border barriers over time also varies widely across countries, with reductions in only about two-thirds of the countries in our sample.  Countries can be classified into four groups.  The first group, which saw a significant decline in border barriers, includes most of the richer and more developed European economies (Great Britain, Belgium, Holland and Denmark), some rapidly developing European economies (Poland and Hungary), and Korea.  The second group, which saw a moderate decline in border barriers, consists of the second tier of European economies in terms of either size or development (Sweden, Spain and Portugal), some economies in transition (Slovenia and Slovakia), and Canada and Japan.  The third group, which saw an increase in border barriers, includes the less developed economies larger economies (Greece, Estonia, Latvia and Lithuania).  Finally, Germany and the US are considered separately because even though we see an increase in the border barriers for these economies, the level of their barriers is the lowest in both the initial and final year of the sample. 

In order to confirm the link between the effects of globalisation and country size, we split the countries in our sample into quintiles according to their real GDP and we plot the evolution of border barriers (Figure 3; as before Germany and the US are placed in a separate group).  It is evident that the smallest countries face the highest barriers and these barriers have remained stable.  The border estimates for the four upper quartiles decrease over time and there is evidence to suggest convergence among the upper quartiles.  Thus globalisation effects (in this sense) are convergent within a set of larger economies and divergent between the smallest countries and the rest.  Underneath this overall pattern, globalisation forces play out in different ways for individual services sectors.  For instance, the size of border barriers appears to converge between small and large economies in transportation but diverges in sectors such as finance, insurance and R&D.

Figure 3. Globalisation and services trade

On the determinants of the borders in services trade

Given the magnitude of estimated border effects, an obvious question is what determines these barriers.  Structural gravity theory suggests that border barriers consist of three principal components: country-specific internal trade costs, country-specific border barriers, and an average (across countries) border effect.  Thus, guided by theory and capitalising on the new multi-dimensional data set of border estimates, we examine the role of domestic institutions, geography, size, and digital infrastructure as determinants of border barriers to services trade.  For instance, we find that internal distance lowers inferred border barriers as it raises internal trade costs.  Conversely, business-friendly domestic regulations that lower internal trade costs result in higher inferred border barriers.  We also find evidence for the positive effect on pure border-crossing costs of advanced digital infrastructure, which facilitates services trade and thus is associated with lower border barriers.  These results offer insights on factors determining the size of border effects, some of which are potentially amenable to policy reform and, therefore, unveil channels that may translate unilateral policy intervention at the national level into changes in the volume of international services trade.

Conclusion and implications

The heterogeneous border estimates that we obtain partly emanate from exogenous characteristics such as geography, but also from institutional features such as the rule of law and private sector development policies, as well as from the quality of digital infrastructure.  As some of these areas are potentially amenable to policy reform, the findings thus point to areas for effective policy intervention.  For the smallest economies, such reforms could offer a way of mitigating the divergent forces that underpin the apparent stagnation of their border barriers. 

It should be noted, though, that the small economies for which we have mapped out border effects still are developed countries, and it will be interesting to see what patterns emerge once we look beyond the OECD membership.  At present, data deficiencies in services would seem to prohibit comparable estimation of border barriers for a wider set of countries.  But the successful decomposition of border barriers according to their structural components provides a way forward.  We develop a projection method that can, in principle, generate the required data, thereby facilitating the estimation of trade costs in services for economies beyond the developed country realm.  While the current analysis in our study focuses on services trade, our methods can be applied similarly to goods trade with potentially large payoffs.


Anderson J E, I Borchert, A Mattoo and Y V Yotov, (2015), "Dark Costs, Missing Data: Shedding Some Light on Services Trade," NBER Working Paper 21546

Francois, J and B Hoekman (2010), “Services Trade and Policy”, Journal of Economic Literature 48 (3): 642–692.

Jensen, B and L Kletzer (2005), “Tradable Services: Understanding the Scope and Impact of Services Outsourcing”, Institute for International Economics Working Paper 05-9.

United Nations (2015), “World Economic Situation and Prospects: Mid-2015 Update”.

Waugh, M E (2010), “International trade and income differences”, American Economic Review 100 (5): 2093–2124.


1 We confirm that the gravity model works well with services trade. In addition, we also document some important differences between goods and services trade.

2 The negative value should not be taken literally. The border estimates are all relative to a global average effect of internal to international trade that is not identified by the gravity model. The missing scaling term is presumably positive, shifting the entire set of data points upward in Figure 1. The US case also illustrates that what we call border barriers may represent differences in tastes; US residents may have an unusual desire to travel abroad, all else equal.



Topics:  International trade

Tags:  services, trade costs, Border effect

William B. Neenan Millenium Professor of Economics, Boston College

Lecturer, University of Sussex

Research Manager, Trade and Integration, World Bank

Assistant Professor of Economics and International Business, LeBow College of Business, Drexel University