Expert Knowledge in Global Trade - Big (Bad?) Numbers: Computable General Equilibrium Modeling and Trade Policy

By Clive George - 08 February 2016
Big (Bad?) Numbers: Computable General Equilibrium Modeling and Trade Policy

Clive George explores the use, misuse and abuse of numbers through computable general equilibrium (CGE) models in predicting the gains from trade agreements.

In promoting the proposed Transatlantic Trade and Investment Partnership (TTIP) between the European Union and the United States, Europe’s Trade Commissioner Karel De Gucht claimed that it would boost growth in the EU and US to the tune of 0.5 percent. If that were true it would be quite impressive, but in failing to understand the numbers provided by his economists, Commissioner De Gucht overstated his case by at least a factor of ten, an error that was then propagated by the BBC. The figure of 0.5 percent on which the Commissioner’s claim was based was not for a rise in growth, but for a rise in GDP, over ten years, boosting annual growth in that period by only 0.05 percent. This misuse of numbers is symptomatic of a larger problem concerning our reliance on computable general equilibrium (CGE) models to grease political tracks or solve policy problems.

The models that are used involve many simplifying assumptions and approximations, some of which are highly questionable, but even if we assume the correctness of the model, the gains from trade that they predict are still relatively small. Recent discussions about the TTIP and the Transpacific Partnership Agreement (TPP) draw attention to the multi-billions of dollars at stake, without mentioning the multi-multi-multi-billion dollar size of the economies involved. By comparison, the predicted gain is only a fraction of a percent. Even this can be an over-estimate. For the EU-US trade deal the numbers quoted were for the most optimistic of the scenarios used in the model. The least optimistic (and perhaps more realistic) scenario gave projections of one-fifth the size of those quoted, i.e. a boost to economic growth not of 0.5 percent, nor 0.05 percent, but 0.01 percent. To put this in perspective, before the 2008 crisis, the US and European economies both grew at over 2.5 percent a year for more than three decades, amounting to about 30 percent in each decade (1). According to the EC’s CGE model, the TTIP would boost the US and EU economies, at most, by a mere 0.5 percent in a decade. If the United States and Europe are to restore their previous rates of economic growth, where are they going to get the other 29.5 percent from? According to the numbers produced by the EC’s own economic model, the “biggest bilateral trade deal in history” would need to be repeated 60 times to get us back to where we were. Yet political excitement for the TTIP remains undampened.

Given the questionable use of numbers, why do politicians appear to be so reliant on them? Are they trying to make a bitter pill easier to swallow? Or is it that these superficially attractive numbers are ripe for the picking, whatever their validity? The two major problems explored in this post—with the model itself (and the data it relies on) and interpretation of the results—are compounded by blind faith in economic expertise. The numbers generated by the models hide conflicts rather than solving them, and mask the politics (and political motivations) under a cloak of apparent inevitability.

In general, we trust the numbers we are given. When the pump says it has delivered twenty liters of petrol we believe it. When politicians promise billions of dollars in gains from free trade agreements (as David Cameron has done with the TTIP) we trust them too. And we trust the people who created the models that produced the numbers. Despite catastrophes like the 2008 financial crisis, economists are widely trusted, along with other experts, for the authoritative and legitimate knowledge they hold. This is, in part, because they make use of quantitative models that appear to produce ‘objective’ truths.

Some trade models can be extremely useful, particularly in alerting policy makers to the dangers inherent in their proposals. Partial equilibrium models serve this purpose, by providing an estimate of the impact of proposed policy changes on particular economic sectors (such as the motor industry). CGE models can do the same, with less precision, but are used mainly for their ability to predict the overall impact on the economy. However, their use and interpretation can be highly political. Damage to certain domestic sectors may be glossed over in presenting only a general picture. The findings may also be used selectively, for example focusing on exports and ignoring imports. This was the case in the European Commission’s press release on TTIP, which championed the model’s estimate of increased exports while ignoring the increased imports, despite the fact that they were predicted to rise even more. In the case of the China-Australia FTA, this has been dubbed “the export agreement” in the Australian parliament, even though it was predicted to increase exports by a mere 0.5 percent, and imports by five times as much.

The economic gains (or losses) estimated by a CGE model are not necessarily reliable predictions. Equilibrium analysis has been subject to severe criticism since its inception, not least because no real economy is ever in equilibrium. What we call an equilibrium ‘model’ is no more than a set of simultaneous equations that calculate the difference between two supposed equilibrium states. One of these represents the current situation, while the other incorporates the proposed changes in import taxes and other trade parameters. In no sense does the analysis model the actual economic processes through which unneeded production facilities close, new ones are planned and then built, land is put to new uses, and people who become surplus to requirements in one economic sector seek employment in another. The calculation provides only an indication of the eventual outcome in what is referred to as the ‘long-run’, when the imagined equilibrium conditions have been restored. John Maynard Keynes famously pointed out that in the long-run we are all dead.

Although certain dynamic features are sometimes incorporated into the static equilibrium framework, CGE models are incapable of addressing the inherently dynamic nature of economic development. They do not address structural factors such as those which transformed South Korea from one of the poorest countries in the world to among the richest in the space of a few decades. When trade liberalization is accompanied by appropriate development policies it can help to achieve growth rates far in excess of those calculated by CGE models. Conversely, as in countries that suffer from the ‘resource curse’, when liberalization is accompanied by inappropriate policies or no realistic development strategy at all, the impact can again be far larger than the CGE projections, but in the opposite direction.

The limitations of CGE models are no secret. In an audit of the European Commission’s management of trade policy, the European Court of Auditors found that the Commission had not appropriately assessed all the economic effects. It identified fundamental weaknesses in the CGE models—particularly the tenuous analysis of long-run effects—and the datasets used, such that they were only suitable for simulation and not for forecasting. The European Commission acknowledged that the models have limitations, but described them essentially as the best economists can do, being “widely accepted worldwide by virtually all similar international organizations”. In other words, irrespective of their limitations, the use of CGE models for trade policy analysis is solidly entrenched within the mainstream economics profession and in the establishment that uses the findings.

Ironically, although the results of CGE analysis are widely used in efforts to win the public over and to influence the policies of other parties to a trade agreement, they are seldom used to develop a country’s own trade policy. Lobbying by powerful interest groups can be far more influential, in rich countries as well as poor ones. Whether or not lobbying plays a role, a comprehensive strategy to enhance a country’s economic competitiveness, through technological development or other means, can far outweigh the minimal gains predicted by CGE models. The EU’s 2000 Lisbon Strategy for Growth and Jobs aimed to make Europe “the most dynamic and competitive knowledge-based economy in the world,” in response to growth in its competitors, primarily China and India. This strategy informed the subsequent 2006 trade strategy Global Europe: Competing in the World. This made no mention of the numbers generated by trade economics models. Instead, it focused on efforts to persuade low and middle-income countries to remove their remaining barriers to European exports and investment, while granting Europe unrestricted access to their resources of raw materials. This is the archetypal strategy for trading success—buy low, sell high.

So what are the ‘real’ motivations behind the TTIP? Are the EU and the US trying to ward off foreign competitors and maintain economic dominance in a post-hegemonic era? The projected economic gains are small but the agreement is being sold to the public as though they were big. Perhaps the gains that are being sought really are big, but if so it is not for the reasons that are claimed. Those with power are able to frame the debate in a way that limits the realm of what is seen as possible or legitimate. We might not yet know the political motivations behind the TTIP and other proposed trade agreements, but trade politics has nothing to do with equilibrium theory. It has everything to do with economic structures and power relations.

How inevitable is the use of dubious numbers in promoting trade policy? The economics profession is beginning to change, in part because of the financial crisis—which opened up space for heterodox voices—and in part through the efforts of a new generation of economists, who recognize the limits of mathematical modeling, and who value the roots of their discipline in the social sciences, the political sciences, and even, as in the case of Adam Smith, moral philosophy. But this won’t stop politicians seeking evidence to support their policies, nor from masking the true reasons for those policies. So what can be done? Are we condemned to accepting trade agreements such as the TTIP and the TPP on the advice of politicians and the economists they choose to employ? Perhaps the biggest hope of avoiding such a fate is that a combination of legislative inertia and civil society opposition on both sides of the Atlantic will prevent the EU-US negotiations from achieving any more than a face-saving deal with minimal effect. From there it is not hard to imagine a re-constituted World Trade Organization seizing the initiative again, to re-shape the face of international trade agreements in a radically different form.



(1). The historical growth rates are calculated from the inflation-adjusted data in World Bank, “World Development Indicators,” (Washington, DC: World Bank, 2014) At non-adjusted current prices the growth rates were even higher.


Clive George is a retired Senior Research Fellow in the Institute for Development Policy and Management, University of Manchester, and Visiting Professor in the Department of International Relations at the College of Europe, Bruges. He is author of The Truth About Trade (London: Zed Books, 2010) and of numerous academic articles on the economic, social and environmental impacts of international trade agreements.

Disqus comments