Cato Online Forum

The Politics of Modelling Trade Agreements: plus ça change

By Gabriel Siles-Brügge
October 2015


When politicians and trade negotiators speak of the merits of the Transatlantic Trade and Investment Partnership (TTIP), they tend to stress the agreement’s vast economic benefits. UK Prime Minister David Cameron referred to TTIP as a “once-in-a-generation prize,” generating vast amounts of growth and jobs. Given a crisis- and austerity-ridden Europe, this agreement represents, in the words of then EU Trade Commissioner Karel De Gucht, “the cheapest stimulus package you can imagine.” Meanwhile, U.S. President Barack Obama sees the TTIP (and the Trans-Pacific Partnership or “TPP”) as a way to “help the entrepreneurs and small business owners who create most jobs in America.”

Particularly prominent in the European debate over TTIP has been a March 2013 study by the Centre for Economic Policy Research (CEPR), which put a precise figure on the gains to be expected from the deal: extra annual GDP of €120bn for the European Union and €95bn ($120bn) for the United States, or 545€ and 655€ ($830), respectively, per family of four.1 There have, of course, been other econometric studies, predicting both much greater and somewhat smaller gains, but the CEPR study has provoked the most heated debate.

In that debate, many have characterized the CEPR figures as vastly exaggerated, yet fairly insignificant anyway, amounting to just “an extra cup of coffee per person per week.”2 A prominent rival econometric study, however, purports to show that TTIP will instead lead to significant losses in both employment and GDP.3 Meanwhile, others have argued that the modelling techniques underestimate the benefits by not taking into account (above all) the dynamic, productivity-enhancing effects of trade agreements.4

So what are we to make of the modelling of TTIP? Much of the discussion so far has focused on economics — assessing the relative merits of the econometric methodologies. This is an important, but incomplete, exercise, however, because it fails to consider the potential for subsequent exaggeration or misappropriation of the results to skew sentiment toward supporting a trade agreement that is controversial for reasons that are not being measured in the models. There is a history behind this concern. But let us first turn to what the TTIP models predict and on the basis of what assumptions.

How and What Do the TTIP Models Predict?

The recent models of TTIP’s economic effects fall into two categories: those that use standard computable general equilibrium (CGE) modelling (such as the CEPR study) and those that rely on a gravity model (notably the Bertelsmann/ifo5 and Raza et al.6studies).

In the case of the former, the basic parameters of the model are to assume that a state of “general equilibrium” exists in the economy as a whole (following the standard Arrow-Debreu theorem); all supply meets its own demand in product and labor markets (in other words, all goods and services are consumed and there is full employment). The modellers then estimate the effects of a reduction in trade barriers on trade flows (and, hence, on aggregate income) with reference to a baseline scenario of no policy change.

The gravity-based models predict the impact of TTIP on aggregate income the other way around. They estimate what reduction in trade barriers is necessary to generate a particular increase in trade flows, based on data from previous trade agreements (and relying on so-called “gravity equations,” which assume large trade creation effects from trade agreements between geographically proximate and economically large countries).

Unsurprisingly, given that they assume different amounts of liberalization, the existing econometric studies produce a range of estimates of TTIP’s effects, from zero in a Centre d’Etudes Prospectives et d’Informations Internationales (CEPII) study to the famously superlative 4.82 percent increase in GDP for the United States in the Bertelsmann/ifo study. The much-debated CEPR estimate of GDP gains of €120bn for the European Union and €96bn for the United States assumes a 100 percent reduction in tariffs and a 25 percent reduction in non-tariff barriers (NTBs) — barriers arising primarily from differences in regulatory practices across the Atlantic.

The European Commission has touted the CEPR study as the most reliable on the grounds that it is not only “at the mid-range of most other studies,” but because CGE modelling is “state of the art.” Such models are seen as the standard tools for the trade. In contrast, the Bertelsmann/ifo gravity modelling is generally seen as more untested, with relatively few details given in the original study as to the methodology employed.

Managing “Fictional Expectations”

Is this a fair characterisation of the CEPR study? Certainly, it is the case that the modellers have explicitly stated their assumptions within the text of the study, that CGE modelling is widespread in the field of trade, and that it is technically complex. It is also fair to say that CGE modelling techniques (and the computing power of the processors churning out the numbers) have evolved since it was deployed (to great fanfare) to model the economic effects of the North American Free Trade Agreement (NAFTA). But much like a lot of the econometric hot air around NAFTA — the predicted benefits in some studies of GDP gains for Canada, the United States, and Mexico did not materialize — there are also reasons to be skeptical here.

Importantly, there is a political motive to exaggerating the potential benefits of the agreement and downplaying its potential costs. The CEPR study serves the purpose of — to borrow a term from the economic sociologist Jens Beckert — “managing fictional expectations.” Even though the future is inherently unknowable, insofar as the social world is too complex and contingent to predict (especially something like the final terms of a multifaceted trade negotiation), modelling creates the ‘fictional’ impression that outcomes can be neatly predicted and quantified.

But while the model predicts meaningful economic benefits, there are many problematic assumptions in the CEPR study. Does the economy actually operate in “general equilibrium”? The persistence of unemployment and unsold supermarket stock would suggest otherwise. And the assumption that TTIP will eliminate a quarter of all remaining NTBs between the European Union and the United States does not seem too problematic, until we realize that only 50 per cent of NTBs are “actionable,” or within the reach of policy to address (other NTBs, including consumer tastes, are less malleable). Moreover, the manner in which NTBs are calculated (by relying on business surveys and conversations with “sectoral experts”) may also skew the results toward overstatement of the impacts of NTBs on trade costs.

Finally, and most significantly, modelling differences in regulation as non-tariff barriers to trade — in other words as something inherently negative — ignores the potential social costs that may be borne from eliminating these barriers. Would TTIP necessarily lead to increased welfare if it downgrades financial services regulation in the United States or food safety standards in the European Union? The point here is not to say that this is going to be the outcome of TTIP, or that all forms of regulatory convergence are detrimental to social and environmental protection, but rather that this eventuality is downplayed by focusing exclusively on the “assured” economic gains predicted by the studies.

Despite the considerable uncertainties surrounding modelling, which render the associated predictions “fictional expectations,” the CEPR study was presented with great pomp by the European Commission without any meaningful caveats. In a leaked internal document from November 2013, the Commission was even found to be arguing that it was crucial “to define, at this early stage in the negotiations, the terms of the debate by communicating positively about what TTIP is about (i.e. economic gains and global leadership on trade issues).” For their part, the CEPR modellers also buried uncertainty about the model results within the text of the study, rather than meaningfully stating this up front in the executive summary or in the “Key Findings” section. That section only contains a brief statement about the difficulties of eliminating NTBs.

Fighting Fire with Fire

In 2014, a Tufts University study predicted that TTIP would generate income losses of between 165€ and 5000€ per EU worker (depending on the EU member state), and a loss of around 600,000 jobs EU-wide. The econometrics in the study was based on the United Nations Global Policy Model, which differs importantly from the CGE model in that it makes the significant Keynesian assumption that employment is dependent on the level of aggregate demand.  Citations to the results of this study by various NGOs challenging the European Commission’s more rosy narrative were subsequently criticized by certain TTIP supporters as “mercantilist” and for reducing trade liberalisation to a zero-sum game.7

But it is important to recognize that many of those who are critical of TTIP are essentially “fighting fire with fire.”  Like supporters of the CEPR study findings, the skeptics are “managing fictional expectations” by endorsing estimates of the effects of the yet-to-be-finalized agreement, as though they were established facts. More significantly, however, they are accepting the terms of the debate as set by the CEPR study, which reduces TTIP’s value to an inherently distributional question. How many jobs will it generate? How much will it boost GDP? Rather than asking what the agreement’s wider impact might be on the likes of public health (e.g., by leading to increases in the consumption of alcohol and unhealthy foods) or the environment (e.g., by reducing levels of protection), those citing the Tufts study are implicitly accepting the logic of those who refer to the agreement’s critics as anti-growth and anti-jobs.8


One reading of this essay might be that if economics is the dismal science, then econometrics is the dishonest science. But despite identifying some of the limitations on modelling trade agreements, the broader point is that the real problem is the use of their results to “manage fictional expectations” — to present with disingenuous clarity the expectations of an uncertain future. The limitations, the warnings, the fine print should get at least as much attention.

In fairness to the European Commission, it has relied less and less on the specific economic figures and more and more on simply touting the economic benefits of the deal more generally, especially for Small and Medium-Sized Enterprises (SMEs). The White House and the U.S. Trade Representative’s office, meanwhile, have been more circumspect about promising specifics in terms of growth and jobs, heeding the lessons of their predecessors, who may have oversold the benefits of NAFTA.

Estimating the benefits of trade agreements with econometric modelling before they are concluded has a long history. What appears to happen rarely is any modelling of the effects of trade liberalisation after agreements come into force. That reinforces the point that establishing a “fictional expectation” of significant gains is more important, politically, than is rigorously evaluating the impacts of trade agreements. TTIP is only the latest example.

1 Center for Economic Policy Research, “Reducing Transatlantic Barriers to Trade and Investment: An Economic Assessment,” (London: 2013)
2 Moody, G. (2014), “TTIP Update XXI: TTIP is Worth an Extra Cup of Coffee Per Person Per Week,” Computer World UK Blog, 27 March. Available from:
3 Capaldo, J. (2014), “The Trans-Atlantic Trade and Investment Partnership: European Disintegration, Unemployment and Instability,” Global Development and Environment Institute Working Paper No. 14-03. Medford, MA: Tufts University.
4 Bauer, M. and Erixon, F. (2015), ‘ “Splendid Isolation” as Trade Policy: Mercantilism and Crude Keynesianism in the “Capaldo Study” of TTIP’, ECIPE Occasional Paper No. 03/2015, Brussels: ECIPE.
5 Bertelsmann and IFO (2013), ‘Transatlantic Trade and Investment Partnership: Who Benefits from a Free Trade Deal’, Berlin and Munich: Bertelsmann Stiftung and IFO Institute. Available from:
6 Raza, W., Grumiller, J, Taylor, L., Tröster, B., von Arnim, R. (2014), ‘Assess TTIP: Assessing the Claimed Benefits of the Transatlantic Trade and Investment Partnership’. Vienna: Austrian Foundation for Development Research.7 Bauer and Erixon8 Capaldo.

The opinions expressed here are solely those of the author and do not necessarily reflect the views of the Cato Institute. This essay was prepared as part of a special Cato online forum on The Economics, Geopolitics, and Architecture of the Transatlantic Trade and Investment Partnership.

Gabriel Siles-Brügge is a professor at the University of Manchester. This essay draws on an argument in his forthcoming book with Ferdi De Ville (Ghent University), TTIP: The Truth about the Transatlantic Trade and Investment Partnership (Cambridge, Polity).