Clarifying my take on dueling trade models

February 2nd, 2016 at 3:38 pm

[See “update” at end of post–JB]

The NYT reported on this presentation yesterday at which I spoke on the results from a new trade model, by economists from Tufts University, of long-run growth outcomes from the TPP.

The Times piece characterized me as there to “support the authors.” That’s correct. But I want to be very clear about the difference between supporting the authors and believing their model. My comments at the session were clear on this point, but in the interest of personal consistency and smart policy making, let me reiterate here. I think my points were nuanced, perhaps too much so in the hurly-burly of this often unfortunate trade debate.

The argument is between dueling models of the TPP’s growth effects. The Peterson Institute (PI) touts a “CGE” model; the Tufts (T) folks use a different model. Their paper–linked above–does a good job explaining the differences in approach and results. The PI model predicts that US GDP will be 0.5% higher by 2030. The Tufts model predicts it will be about that much lower after 10 years.

I don’t believe either result. That is, as I stressed in my comments, the idea that we can reliably predict an impact of this magnitude, 10-15 years from now, from a 12-country, 6,000 page trade deal, should not be taken even the slightest bit seriously (better yet, the PI model predicts 0.1% growth from the TPP–$1 billion–by 2020). The authors of both studies should be very clear about this point–such quantitatively small impacts cannot be reliably distinguished from zero impact.

So why was I there? Well, first to make that point. If the policy community and the press want to make an informed decision about the TPP, they should not do so based on these models. I’m somewhat shocked that more US economists–including those who generally support trade expansion–are not saying the same thing (FTR: Dani Rodrik excepted: “[the models] give us a sense of precision that really doesn’t belong there.”)

Second, to point out that the T model at least makes more reasonable assumptions than the PI model. The latter assumes full employment, balanced trade, and no changes in the distribution of wages and profits.

Dean Baker’s been hammering this balanced trade point, quoting from the PI’s study: “The model assumes that the TPP will affect neither total employment nor the national savings (or equivalently trade balances) of countries.” He writes:

In the wake of the Great Recession many of the world [sic] most prominent economists (e.g. Larry Summers, Paul Krugman, Olivier Blanchard) no longer believe that the economy will automatically bounce back to full employment. They now accept the idea of “secular stagnation,” which means that economies can suffer from long periods of inadequate demand. If secular stagnation is a real problem, then there is no basis for assuming that the demand and jobs lost due to a larger trade deficit can be offset by other policies.

The T study relaxes these assumptions, and it makes a difference in their findings. That doesn’t mean they’re right. As I’ve said, “right” when it comes to the TPPs impact on growth 5, 10, 15 years from now is beyond our modelling capability. But I was there to support their effort to at least try to see what happens when the modelling exercise plugs in more realistic assumptions. And as economist Josh Bivens likes to point out, these models’ results are hugely driven by their assumptions.

So how do we decide whether or not to support the TPP? We do the hard work of rolling up our sleeves and seeing what’s in and what’s not in the beast. I’ve been trying to do so, and below, you can see my latest take, pasted in from testimony I just gave this AM before the Ways and Means committee (note: CBPP does not generally take a position on trade policy).

As you will see, I do not oppose the agreement. These trade agreements are no more than “rules and the road” for countries engaging in trade, and such rules can be useful. Or they can be damaging, both to our own interests and those of emerging economies (e.g., patent protections which make medicines more expensive). The only useful way forward is to focus on these institutional arrangements, ask who’s being fairly and unfairly represented, and what procedures are being put in place to enforce rules and resolve differences.

The 12-dimensional macro models trying to get us to believe that they’re accurately predicting gains and losses decades out are worse than implausible. They’re a distraction from the serious work.


Given this committee’s role in international trade and trade agreements, I wanted to note a few points and concerns regarding the Trans-Pacific Partnership, or TPP, from the perspective of economic growth and opportunity.

Contrary to simple textbook trade theory, the increase in international trade has not been an unequivocal good for all working families. In fact, more realistic theories of trade are quite clear on the point that trade creates winners and losers, with the latter typically including those thrown into competition with cheaper workers abroad. Still, our highly productive workforce can compete globally, as long as the playing field is not tilted against them.

If the benefits of trade are to be more broadly shared, two things have to happen. First, our trade agreements must be more than handshakes between investors. They must provide workers from all signatory countries with the rights and protections they need to capture some of the benefits of trade. Second, we in the US must be able to lower our large and persistent trade deficits through enforceable rules against currency interventions that give our trading partners an unfair price advantage.

The TPP goes further than past agreements in various ways that could protect workers both here and in other signatory countries from unfair labor and wage practices. For example, the USTR worked out bilateral “consistency plans” with Vietnam, Malaysia, and Brunei that specify ways these countries must change their laws and practices to meet the general obligations in the TPP’s labor chapter. Of course, such provisions underscore the need for stepped up enforcement, an area where the US record has not been strong enough. A nonpartisan Government Accountability Office survey of this issue concluded that “monitoring and enforcement [of labor provisions in prior trade agreements] remain limited.”

An even greater concern is the absence of a consistency plan for Mexico, particularly because US auto production has been sharply increasing there. Mexican workers are typically unable to unionize or collectively bargain, and they make less than a fifth of what US autoworkers are paid. This combination of accelerated outsourcing of auto production to Mexico and suppression of workers’ rights there reduces living standards and increases inequality on both sides of the border.

Thus, in the spirit of trade that is both pro-growth and pro-worker, I urge this committee to carefully consider both enforcement and oversight provisions in the TPP, and the need for a plan to improve labor rights in Mexico.

On currency, the existing side agreement to the TPP has some positive features but no enforcement mechanism. As economist Joe Gagnon points out, the “TPP partners merely reiterate the obligation they already have as members of the International Monetary Fund (IMF) to ‘avoid manipulating exchange rates … to prevent effective balance of payments adjustment or to gain an unfair competitive advantage.’” The side agreement may well provide the information needed to quickly identify currency manipulators, but voluntary agreements only work if key actors, such as those at the US Treasury, take corrective action in the face of evidence. Unfortunately, our history here is lots of evidence and virtually no action. In the face of obvious currency management by China, for example, the US Treasury has been extremely hesitant to label them a currency manipulator.

The absence of a currency chapter in the TPP suggests the need for Congress to legislate enforceable currency rules outside of the trade agreement. For example, back in 2010, this chamber, while no less divided than it is today, overwhelmingly passed legislation that, if it had been enacted, would have allowed the Commerce Department to treat currency management as an unfair subsidy, calling for countervailing duties. Given the long history of voluntary measures being inadequate to the task of pushing back on currency manipulation, such enforceable rules would be preferable to the voluntary approach.

Other aspects of the TPP also warrant close scrutiny. The fact that investors are using the investment dispute settlement procedure under NAFTA to challenge the administration’s decision on the Keystone pipeline underscores the importance of making sure our sovereign rights are adequately protected. The agreement also has weaker rules of origin for automotive products than past trade agreements (e.g., NAFTA), which could hurt employment opportunities along our supply chains for cars and car parts.

Update: Jeronim Capaldo, one of the authors of the T study, writes me a note in which he expresses agreement in all of the above. He says, “I find the debates on half percentage points in projections almost grotesque.”

He goes on:

However, as far as I can tell, this very reasonable point does not get a lot of traction in the public debate. My hope is that by offering alternative numbers we can stimulate some interest in what’s underneath the various explanations. The diatribe about +0.5 and -0.5 is ridiculous but the debate about what changes we can expect from TPP or other forms of liberalization seems important to me.

This said, if I may suggest a slightly different perspective, I’d say that both models are equally “accurate” (very little!) but not equally “plausible”.

That last point is well taken, especially since one of the points of my post was that the assumptions in the T model were a lot more realistic than those of the PI model.


Print Friendly, PDF & Email

2 comments in reply to "Clarifying my take on dueling trade models"

  1. Smith says:

    Read this too:
    No to TPP and the candidates who were for it before they were against it.

  2. JohnR says:

    I see this a lot in all sorts of areas- it’s the Curse of Statistics, where people have transferred to mathematical analysis their old love of Oracles and Witches. The perceived accuracy of the stated results, however derived, is all that counts. The ‘slop in the system’, which is far more useful and informative in most cases, is tossed aside. I’m frequently reminded of Alfred Bester’s tribe chanting “Quant Suff!” when I see the uncritical acceptance of potentially meaningless (or rather, not as meaningful as claimed) results. Ackbar’s Warning should be invoked much more often than it is when P-values are reverently placed on the altar, as if they are the Word of God. That being said, of course, math and statistics often tell you extremely useful things that are not intuitively obvious. However, like language, they can be used to hide and mislead (intentionally or not) as easily as they can communicate and reveal.