Do comprehensive trade agreements generate more trade? A conjecture to reconcile alternative theories
Ours is an age of free trade agreements (FTAs) of all sorts and of various depth and breadth in the issues they cover – simple bilateral preferential trade agreements, broad economic partnership agreements, regional trade agreements, customs unions, monetary unions, and single markets. And this only scratches the surface of the tangled web of agreements that cover a range of trade-affecting issues from investment to double taxation to patent office cooperation.
The age of FTAs – nowadays deep and comprehensive FTAs (DCFTAs) – is also the age of global value chains (GVCs) and multinational enterprises (MNEs). The latter claim a large share of global cross-border trade and an equally impressive share of domestic markets through their foreign affiliate sales.
MNEs emerged first, GVCs second, and DCFTAs third. Indeed, by the time the age of the DCFTA had arrived in the mid-2000s, the trend of global trade growing much faster than global GDP, which was driven by the formation of GVCs, was maturing. After the 2008-09 global financial crisis, trade as a share of GDP flat-lined following an initial rebound.
Figure 1: Trade agreements over time (by class)
Source: Osnago, Alberto, Nadia Rocha, and Michele Ruta. 2016. “Deep Agreements and Global Value Chains.” Global Value Chain Development Report 2016 Background Paper presented in Beijing, China, 17-18 March.
Figure 2: World trade to GDP ratio (1995-2014)
Source: World Trade Organization. 2015. International Trade Statistics.
If timing can be used to infer causality, the era of strong growth in trade driven by intermediates goods trade and GVC formation was associated primarily with shallow FTAs. DCFTAs emerged later, when the trend was established. And, one may infer accordingly, they emerged in an attempt to sustain GVCs in the face of headwinds, such as the increasing thickness of borders due to security measures and the recognition of risks associated with extended supply chains.
Yet, empirical analysis of the impact of FTAs suggests that DCFTAs generate more trade than shallow FTAs. Whether DCFTAs generate more trade – and by extension greater income and welfare gains – is an important issue in the evaluation of agreements like the Trans-Pacific Partnership (TPP).
Two workhorse tools have figured prominently in the effort to quantify the impacts of trade agreements: computable general equilibrium (CGE) models and gravity models. The former simulate the impact of trade agreements as a counterfactual, based on the quantifiable liberalising commitments made in FTAs; the latter seek to identify the ex post impacts of FTAs, based on observed changes in trade patterns.
The general consensus from the gravity modelling literature is that FTAs have a large impact on bilateral trade – potentially doubling bilateral trade after 10 years – and this is primarily due to trade creation.
What can be inferred from the CGE modelling literature, on the other hand, is that FTAs have modest impacts on trade and generate a significant degree of trade diversion through the tariff preferences.
Efforts to reconcile the difference focus on non-tariff barriers (NTBs) to cross-border trade in goods and services and comparable barriers to foreign direct investment. Reconciliation requires large reductions in trade costs from FTAs for both goods and services as well as a reduction in costs/risks for investment.
However, such trade/investment cost reductions are hard to identify in FTA texts, at least if we associate cost reductions with binding, legally-enforceable provisions that are WTO-plus (beyond WTO commitments) or WTO-extra (commitments in new areas). Horn et al., for instance, found a significant amount of “legal inflation” (commitments that are not binding or enforceable) in trade agreements, and that the number of legally-enforceable WTO-extra provisions contained in the EU and US agreements at that point was, in fact, quite small and mainly addressed competition policy in the former and environmental and labour standards in the latter. Moreover, the empirical evidence on the impact of mutual recognition agreements for conformity assessment – the main tool for trade cost reduction in FTAs – is quite disappointing.
The impact of DCFTAs on trade and investment remains, therefore, an open and important issue.
Considered as a narrative, DCFTAs contributed little to GVC formation and are a latecomer in GVC facilitation, emerging when these were more likely to be in retrenchment than expansion. This suggests that the main role of DCFTAs has been to optimise the operating conditions for existing MNE trade, rather than to create new trade, the expansion of which was accomplished by traditional shallow FTAs, which simply eliminated tariffs on industrial goods and improved upon core border-related measures covered by customs cooperation and related arrangements.
Considered in terms of their impact on legal frameworks, recent DCFTAs appear to add few legally-binding commitments that reduce trade costs; indeed, they are likely to be trailing the cutting edge of work in the various areas impacting on trade costs. This falls under the purview of customs officials engaged in their own international networks as well as networks of experts in various fields.
How then do we square the results of ex post quantitative analyses, which suggest that DCFTAs have a larger impact on trade than shallow FTAs, with the evaluation of DCFTA texts, which fails to find the legal/regulatory reforms that could plausibly generate such larger impacts?
I offer a straightforward conjecture to reconcile these findings: DCFTAs have historically liberalised more deeply in traditional areas, including in sensitive sectors and core NTB areas, than did shallow agreements. These effects were attributable to deeper core chapters, not to the breadth furnished by the additional chapters.
This suggests that what we see in trade agreements such as the TPP is what we get – that is, we can evaluate them on the basis of the liberalisation quotients for core measures that are captured in the recently developed indexes of NTBs affecting trade in goods and services and investment.
In the TPP’s case, this is not large. The TPP is essentially a North-North FTA – a Japan-US FTA with an entourage. It does not liberalise deeply in sensitive sectors, its core provisions for services are only partly binding (witness the escape clauses for services sector regulation, which reserve the right for many of the TPP parties to revert to General Agreement on Trade in Services (GATS) bound commitments), and its core provisions for goods trade costs are largely redundant to commitments already made by the parties under the WTO’s Trade Facilitation Agreement. A strict reading of the TPP text fails to find additional significant cost reductions. Other interpretations find larger effects.
How we translate the text into trade determines the bottom line impact. Differences in impact evaluations, in turn, play into the policy discussions. It is important to be aware of these methodological issues underlying the differences in estimates.
This article is based on the discussion paper Do Comprehensive Trade Agreements Generate More Trade? A Conjecture to Reconcile Alternative Theories by Dan Ciuriak.
Dan Ciuriak is Director and Principal, Ciuriak Consulting Inc. (Ottawa), Fellow-in-Residence with the C.D. Howe Institute (Toronto), Senior Fellow with the Centre for International Governance Innovation (Waterloo), and Associate with BKP Development Research & Consulting GmbH (Munich). He is a member of the E15 Expert Group on Reinvigorating Manufacturing: New Industrial Policy and the Trade System.