The world trade system as governed by the WTO works on two fundamental principles, i.e. most-favoured nation treatment and national treatment.
The former means that no WTO member's trade flows shall be treated worse (i.e. taxed) at the border than those of the most-favoured nation. The latter means that – after having crossed the border – goods shall be treated as if they were of national origin. In general, preferring one country's goods over the ones of another country in tariff treatment would violate WTO law. However, the WTO grants an exception from this under the umbrella of preferential agreements which have to be notified to the WTO. Countries may abolish tariffs on individual products or even all products through free-trade agreements or customs unions. This is one legal form of preferential market access, even though it discriminates between goods from trade agreement members and non-members within the WTO. Similar agreements exist for services trade and foreign direct investment (through trade agreements, bilateral investment treaties, and bilateral tax treaties). Let us refer to all such agreements (and, eventually, even to currency unions and currency pegs) as preferential economic-integration agreements, or PEIAs.
Preferential market access in academia and in the real world
Academic economists tend to apply what we call a unimodal approach to business activity and PEIAs (if not economic policy in general). This means that only PEIAs targeted towards goods trade are assumed to have goods trade effects, and similarly for services trade and foreign direct investment. Reality is more complex than that, in particular since the 1990s (see Baldwin 2011a,b, for a host of reasons for this change). First of all, we know now that the biggest firms in the world are big in everything they do, i.e. goods trade, services trade, foreign direct investment, value added, employment, etc., and they account for a discretely large fraction of world activity in all domains. This creates interdependence between the activities we see, and it makes an independent analysis of goods trade or services trade potentially problematic. Second, and probably related to that matter, we see a lot of simultaneity in the signing and enforcement of PEIAs as well as in many other economic policy undertakings. Figure 1 illustrates the evolvement of different PEIA types between 1960 and 2005. It shows a particular surge in the signing of PEIAs on goods and services trade as well as on investment and taxation from 1990 onwards.
Figure 1. Number of PEIAs 1960-2005
In new research (Egger and Wamser 2013a) we show that the typical PEIA integration path starts with bilateral investment or tax treaties and adds goods trade agreements only after some time. By way of contrast, academic work tends to focus on goods trade agreements and treats services trade agreements and all types of investment agreements relatively stepmotherly. Moreover, academic work assumes that one type of PEIAs emerges at a time. Consider goods trade, services trade, bilateral investment, bilateral tax, and currency agreements (unions and pegs) as five modes of PEIAs countries can adopt. This gives countries 21 options to combine PEIAs. (Even when disregarding currency agreements, one would face 15 options.)
Figure 2. Number of bilateral country-pairs by number of PEIAs in 2005
Figure 2 illustrates that, for the year 2005, if countries conclude PEIAs at all, most country-pairs concluded one type of agreement, but the likelihood for combining them is also relatively high. About 82% of the country pairs had not concluded any PEIA and the rest had some in place. Of the country-pairs that granted preferential market access in any one of the five modes, 67% did so in a unimodal way, while 33% combined them with other PEIA modes.
Effects of preferential market access
Why should we care? The problem of every unimodal approach is that coexisting PEIAs are disregarded, and analysing the preferential agreements in isolation leads to two sources of bias. First, an attempt to understand the (political and economic) mechanisms at work to stimulate PEIAs may fail when disregarding the interdependence in the decisions to conclude PEIAs. The results in Bergstrand and Egger (2013) as well as ours (Egger and Wamser 2013a) clearly show that the propensity of, say, participating in goods and services trade agreements and in investment and tax agreements are not independent from each other. Second, cross-activity effects of PEIAs in conjunction with correlated PEIAs lead to biased estimates and partial misattribution of the effects to one mode while another one is responsible. How large are the effects of PEIAs? To shed light on this question, let us consider average direct effects on bilateral goods exports. Our results suggest that the impact of participating in a bilateral investment treaty (or a bilateral tax treaty) only on goods exports is about 2.5 (1.6) times higher than participating in a goods trade agreement. Moreover, our results suggest that there are returns on combining different modes of PEIAs with each other. This is illustrated in Figure 3.
Figure 3. Treatment effects at intensive export margin
That figure illustrates two things.
- First, there is a nonlinear relationship between the number of modes of PEIAs concluded and bilateral goods exports.
In the figure, we consider up to four modes of PEIAs, i.e. goods trade agreements, services trade agreements, bilateral investment treaties, and bilateral tax treaties. On the horizontal axes, we display the treatment scope (number of PEIA modes covered) and the control or comparison scope (number of PEIA modes covered), irrespective of specific PEIA combinations. On the vertical axis, we display the log of bilateral goods exports. The tallest blue bar in the back indicates the treatment effect of covering preferential market access through four modes of PEIAs relative to a situation without any preferential market access.
The staircase on the right-hand side in the back indicates how goods exports are affected when first implementing one PEIA (orange), then adding another one (green), a third one (light blue), and a fourth one (dark blue). The figure suggests that the biggest benefit to be had for trade is when liberalising market access in all considered dimensions, since none of the pillars is (significantly) higher than one of the dark blue ones for a treatment scope of four modes with a control scope of between zero and three modes.
These findings are consistent with the following notions.
- First, the average goods trade agreement is fairly shallow (only covering few products).
- Second, multinational firms are important in conducting and controlling goods trade.
The latter leads to a direct dependence of goods trade flows on policy measures affecting foreign direct investment. This evidence suggests that we may have looked at the wrong as well as on too few policies for a long time when focusing on goods trade. The same may be true for other activities such as services trade and foreign direct investment.
Baldwin, Richard E (2011a), "21st century regionalism: Filling the gap between 21st century trade and 20th century trade rules", CEPR Policy Insight 56, Centre for Economic Policy Research, London.
Baldwin, Richard E (2011b), Trade and industrialisation after globalisation's 2nd unbundling: How building and joining supply chain are different and why it matters. NBER Working Paper 17716, National Bureau of Economic Research, Cambridge, MA.
Bergstrand, Jeffrey H and Peter H Egger (2013), The determinants of BITs, unpublished manuscript.
Egger, Peter H and Georg Wamser (2013a), "Multiple faces of preferential market access: their causes and consequences" Economic Policy 28(723), 145-187.
Egger, Peter H and Georg Wamser (2013b), Effects of the endogenous scope of preferentialism on international goods trade. CESifo Working Paper No. 4208.