VoxEU Column International trade

The urban legend: Pre-GATT tariffs of 40%

Accurately establishing the GATT’s starting point is important for assessments of its contributions to the post-war trading system, on which much of contemporary trade policy continues to be based. However, a frequently cited statistic is that average tariffs immediately prior to the first GATT negotiations in Geneva in 1947 were at or above 40%.  This column largely debunks the 40% myth and suggests that average tariffs in 1947 were around 22%. 

Who was the first to claim that average tariffs were 40% immediately prior to the negotiations of the General Agreement on Tariffs and Trade (GATT) in 1947? Like all good urban legends, pinning down its exact origin has proven difficult.

The source and spread of the 40% myth

Public reference to 40% tariffs in 1947 can be traced at least as far the mid-1980s. The New York Times mentions it in an article covering the Punta del Este launch of the Uruguay Round of GATT negotiations in September 1986 (Farnsworth, 1986). More substantively, the 40% figure can be found in the World Bank’s flagship World Development Report 1987, released in June 1987, which stated:

From the end of World War II until 1974, protectionism seemed to be in decline. Successive rounds of negotiations in the GATT had cut tariffs on trade in manufactures – from an average level of 40% in 1947 to between 6% and 8% for most of the industrial countries – even before the last round of multilateral trade negotiations (the Tokyo Round, 1974-79) had taken place. (World Bank 1987, 134-35)

Nevertheless, and equally important, none of these references points to any data or methodology that would confirm the authenticity of the 40% number or explain its derivation. 

By the mid-1990s, the 40% number had begun to spread like wildfire. It became a standard talking point in statements made by WTO officials; by Democratic and then Republican US Trade Representatives; in testimony by trade officials before the US Congress; and in official reports by the US Government Accountability Office.1 While in 2007 the WTO Secretariat’s economic research staff ultimately pushed back against the 40%, 2 by then it may have been too late. The 40% had seemingly become ubiquitous, appearing in the leading undergraduate Principles of Economics textbook (Mankiw 2004, p. 193), Encyclopedia Britannica, and today it can even be found on Wikipedia.

Why are the 1947 average tariffs so difficult to pin down?

It is worth clarifying up front that it is not as if to the 40% urban legend diverted attention away from some other more plausible average tariff level that was now being overlooked. Our scouring of the historical record suggests the opposite. The 40% number may have arisen simply to fill a void; previous to 40%, no other recognised number seemed to exist.

What makes it difficult to come up with such an estimate? Our recent research (Bown and Irwin 2015) begins by exploring the many contributing explanations behind why any attempt will be fraught with complications:

  1. Data availability: There was no WTO (or GATT) Secretariat in 1947; aside from the US, governments have been much less transparent about their official records on applied tariffs.
  2. The multiplicity of methods for constructing average tariff measures. Confusion can arise with multiple approaches. While each approach has its pros and cons – including feasibility (data availability) – nevertheless, each methodology can have a different implication for interpretation.
  3. The least data-intensive method and downward bias. Constructing a trade-weighted average tariff series only requires information on import customs revenue and the total value of imports, but under-estimates are a noted problem with such a measure.3
  4. The lack of a common product classification system in the 1940s (through until 1988). A potentially more informative series – such as simple average tariffs – may simply turn out to be uninformative for this period for purposes of cross-country or inter-temporal comparisons.

‘Backcast’ estimates for average import tariffs at the GATT’s starting point in 1947

Given these difficulties, and as a starting point to filling the void, we take a different approach.

We begin with the average tariff data at the beginning of the Kennedy Round in 1964; by then, more credible simple average tariffs were being compiled and circulated. Second, we combine that with other credible information on percent average tariff reductions that had resulted from the prior GATT rounds that were also readily available, even though the average tariff levels in each round were not. We use these two elements and simple math to ‘backcast’ (as opposed to forecast) what the 1947 average tariffs would have been to yield the starting point tariffs for the Kennedy Round.

Figure 1 shows our results. Our estimates are that the 1947 average tariffs – i.e. those applied immediately prior to the onset of the first GATT negotiations in Geneva – were about 22%. Notably this estimate is much lower than the 40% urban legend. It also has the benefit of implying a much more gradual time path of average tariff cuts over the full history of the GATT (1947-1994) than would be implied by an initial 1947 level of 40%.

Figure 1. Path of average tariffs: Pre-Geneva to post-Uruguay Round

Source: Bown and Irwin (2015), Figure 1. Constructed by the authors, based on average tariff levels for US, EU and Japan.
Notes: Backcast estimate of pre-GATT 1947 average tariff level of 21.8% based on upper bound assumption of 26% tariff cut in the first (Geneva) Round. Assuming a 21% tariff cut in the first (Geneva) Round would imply an even lower backcast estimate of pre-GATT 1947 average tariff level of 20.5%.

Why does pinning down the pre-GATT starting point matter?

An accurate estimate of the 1947 average applied tariff is important for assessing whether applied tariff cuts that subsequently followed were an important component to the GATT’s overall contribution to the post-war trading system. Contemporary economics research has made increasing strides toward understanding the role and value of trade agreements such as the GATT/WTO (for a survey, see Bagwell et al. forthcoming); as such, recent research is also now beginning to seek to quantify the early GATT’s impacts. This ranges from micro-level studies of early GATT bargaining rounds (Bagwell et al. 2015) to macro-level quantitative studies that compare a world with and without the GATT/WTO (Ossa 2014).

This leads us to our last point, which is the contemporary policy relevance of the early GATT. Aside from facilitating negotiations to reduce applied tariffs, the GATT established a new, multilateral system that continues to serve as the foundation of trade policy today. While these other early contributions are more difficult to capture with simple summary statistics, nevertheless, the 1947 GATT established a system based on rules of non-discrimination (MFN and national treatment), it reduced policy uncertainty by introducing tariff binding commitments, and it created a dispute resolution forum to prevent trade spats from escalating into trade wars.

The open question is the extent to which these fundamental GATT/WTO-provided benefits will withstand the recent shift in emphasis toward the mega-regional agreements embodied in the Trans-Pacific Partnership (TPP) and Transatlantic Trade and Investment Partnership (TTIP).


Aldonas, Grant D. (2001) “Testimony of Under Secretary of Commerce for International Trade Grant D. Aldonas: ‘International Trade Administration: The Commerce Department's Trade Policy Agenda,’” U.S. House of Representatives, Committee on International Relations, June 21.

Bagwell, K., C. P. Bown, and R. W. Staiger (forthcoming) “Is the WTO Passé?” Journal of Economic Literature.

Bagwell, Kyle, Robert W. Staiger and Ali Yurukoglu (2015) “Multilateral Trade Bargaining: A First Look at the GATT Bargaining Records,” NBER Working Paper, No. 21488, August.

Barshefsky, Charlene (1997) “U.S. Trade Policy Objectives and Initiatives,” Hearing before the Subcommittee on Trade of the Committee on Ways and Means, House of Representatives (Serial 105-53), March 18.

Bown, Chad P. and Douglas A. Irwin (2015) “The GATT’s Starting Point: Tariff Levels circa 1947,” CEPR Discussion Paper 10979.

Farnsworth, Clyde H. (1986) “GATT Talks Facing Tough Obstacles,” New York Times, September 22.

Government Accountability Office (1994) “General Agreement on Tariffs and Trade: Uruguay Round Final Act Should Produce Overall U.S. Economic Gains (Volume 2),” Report Number GGD-94-83B, July 29.

Mankiw, N. Gregory (2004) Principles of Economics (3rd edition). Ohio: Thomson South-Western.

Ossa, Ralph (2014) “Trade Wars and Trade Talks with Data,” American Economic Review 104(12): 4104-46.

World Bank (1987) World Development Report, 1987. Washington, D.C.: World Bank.

WTO (1995) “Statement of Ambassador Dr. Mounir Zahran, Chairman of the GATT 1947 contracting parties to the closing session,” Geneva, December 12.

WTO (2007) World Trade Report: Six Decades of Multilateral Cooperation, What Have we Learnt? Geneva: WTO.

Zoellick, Robert B. (2002) “So What Is There to Cover? Globalization, Politics, and the U.S. Trade Strategy,” Address to the Society of American Business Editors and Writers, Phoenix, AZ, April 30.


1 See, respectively, WTO (1995), Zoellick (2002), Barshefsky (1997), Aldonas (2001), and GAO (1994).

2 The WTO’s annual research flagship World Trade Report 2007 suggested average tariffs at the time of the GATT’s founding were “situated in a range between 20 and 30 per cent,” (WTO 2007, p. 207).

3 Import-weighting puts little weight (low imports) on high-tariff goods. The extreme example is a prohibitive tariff – one that is so high, imports for associated products are zero and thus are not included (mathematically) in the average tariff calculation.

3,570 Reads