In a paper called “TTIP: European Disintegration, Unemployment and Instability”, economist Jeronim Capaldo argues that there are flaws in four prominent studies on the effects of the proposed TTIP agreement between the U.S. and the European Union. The problem is two-fold. First, all studies use similar models and data, which means that they all share the same set of assumptions and should thus not be treated as independently reaching similar conclusions:

Methodologically, the similarities among the four studies are striking. While all use World Bank-style Computable General Equilibrium (CGE) models, the first two studies also use exactly the same CGE. The specific CGE they use is called the Global Trade Analysis Project (GTAP), developed by researchers at Purdue University. All but Bertelsmann use a version of the same database (again from GTAP).

A detailed discussion of the shared heritage of the different CGE models can be found in a paper by Werner Raza and colleagues (pp. 37-49), which Capaldo cites.

He then goes one step further and alleges that the underlying econometric models are simply false, or at least inappropriate. According to him, CGE models rely on several flawed assumptions:

• High labor mobility is supposed to allow workers in less competitive industries to switch to those that benefit from trade liberalization, which are assumed to grow enough to absorb the new workforce.
• Overall, the gains for workers with the right skills are supposed to outweigh the losses for others.
• The model assumes that new trade between countries/regions is created (rather than diverted from elsewhere, which would be a zero-sum result).

While I have no training in economics and don’t know the econometrics literature, I realize that all large-scale models of social science need to rely on simplified assumptions. Nevertheless, it seems to me that Capaldo has a point. If his account is correct, then European policymakers should look for more diverse academic input. More generally, if the most widely used models really are blind to potential downsides for labor, then that goes against the interest of European citizens. (As they ought to be very loss averse when it comes to employment as well as skeptical about the distribution of pay-offs from economic gains.)

So how do we come up with an estimate that pays more attention to potential negative effects? Capaldo uses the UN Global Policy Model (GPM), which models economic activity as demand-driven, explicitly models different regions, and includes an estimate of employment. (Again, I lack the knowledge to assess how this works and how much sense it makes.) In this model, unemployment and household income are projected to deteriorate in the long term (2025) for several European countries, as aggregate demand is lowered due to trade diversion (see pp. 10-19 for this and other findings).

Capaldo is pretty transparent about the limitations of this approach:

• the non-TTIP baseline scenario (which serves as a comparison) might be wrong
• the chosen model might be as ill-specified as the ones he is criticizing
• policy responses down the road are not included (and that’s hardly possible)
• …and the paper completely ignores the investment dimension of TTIP (which is a weakness shared by the CGE models, according to Raza et al., p. 49)

So the headline “TTIP will lead to a loss of 600,000 jobs” does not really do the paper justice, although the author himself uses pretty strong language in the conclusion.

No matter which forecast turns out to be better in the end, this discussion shows that policy decisions should not rely on a single strand of academic analysis. There is a lot of uncertainty involved in these negotiations, and I don’t see how there can be a confident forecast of net effects.

One final note for the political debate in general: TTIP opponents should not forget that the status quo will not necessarily be maintained or improved just by inaction. The people likely to lose from TTIP are probably heading for difficult times anyway, leading to questions about how to compensate them. Whether European leaders will decide in favor or against TTIP, they are making high-stakes bets on how globalization will play out over the next decades.

Thanks to Zoe for pointing me to the study. And if anyone can add insight regarding the comparison between the different models, please let me know!