Stanford Responds to Moffatt: Why I Still Worry About Auto Job Losses Under a TPP

My friend and fellow #cdnecon tweeter Mike Moffatt has published a thought-provoking commentary regarding the impact of the proposed Trans Pacific Partnership (TPP) on Canada’s auto industry. Specifically, Mike engages critically with previous arguments I have made (on this site and elsewhere) that the TPP, as currently negotiated, could result in the ultimate loss of tens of thousands of Canadian auto jobs, due to the relocation of a significant proportion of the industry’s North American supply chain. In this blog post I will respond to his three main arguments.

I should start with a small point of numerical clarification. Mike’s critique refers to the first incarnation of my TPP auto job loss estimates, developed when the negotiations were nearing their conclusion last summer. At that point, media leaks indicated that the draft text (originally developed by Japanese and U.S. negotiators, without the participation or even knowledge of Canadian and Mexican officials) would include a big reduction in regional value thresholds from 62.5% to 45% for finished vehicles, and from 60% to 30% for auto parts. Based on the simulation methodology described in my initial blog (and summarized accurately by Mike in his commentary), this would permit a decrease of some 24 percentage points in the weighted average content threshold (equaling the weighted average of a 17.5-point reduction for finished vehicles and a 30-point reduction for parts). On that basis I originally calculated a potential loss of 24,000 auto jobs, representing the proportionate loss in Canadian auto manufacturing employment resulting from the relocation offshore of that proportion of total value-added production.

Strong concern over the negative effects of that initial proposed deal was expressed by Canadian and Mexican negotiators, and in the final negotiations (in Atlanta in October) the value threshold for auto parts was marginally strengthened (it is now between 35 and 45%, with different thresholds applied to different types of parts). That change reduces the weighted average reduction in the combined automotive threshold to something under 20 percentage points, and hence the job loss estimate is reduced accordingly to around 20,000 positions. The last-minute change in the TPP’s auto provisions explains the difference between the original 24,000 job loss estimate and the 20,000 estimate which I (and Unifor) have been referring to since the final text was negotiated.

Keep in mind, as expressed in my initial commentary, that this is a long-run potential effect resulting from the geographical restructuring of the North American auto supply chain, as would be permitted by the weaker TPP regional content rules. It would occur over several years on the assumption that automakers and top-tier suppliers adjust their sourcing decisions to take advantage of lower-cost offshore components (yet still preserving their tariff-free status within North America). Keep in mind that there are other potential avenues for automotive job loss arising from the TPP, as well: including jobs lost as a result of a wider trade automotive imbalance with Japan (an almost certain result of the TPP, given that Japan has no auto tariff to eliminate, and hence bilateral tariff elimination will give their sales in Canada a boost with no impact on Canadian exports going the other way), and potential fall-out from the asymmetrical timeline for auto tariff elimination (Canada’s would be gone in 5 years, while U.S. vehicle tariffs would take up to 30 years to disappear an enormous and arbitrary difference that will undermine efforts to attract new investment into Canadian auto plants).

Now let’s get to the three main objections which Mike makes to my reasoning:

1. “What is the counterfactual?” Mike suggests there is no point comparing a TPP scenario to the current status quo case, since the TPP is going to happen with or without Canada. The more important comparison is between a TPP involving Canada, and one excluding Canada.

I disagree with this assumption. To be sure, Canada cannot dictate the terms of a 12-country deal. But neither are we an idle bystander, only able to take or leave what the others come up with. Canada (under the Harper government) played an active role in pushing the TPP talks to their flawed conclusion in Atlanta in October. And by indicating publicly (on national TV, no less!) that Canada had no choice but to sign a TPP, no matter what harm it did to our largest export industry, former PM Harper squandered our national bargaining power in those final stages of the talks. Even today the deal is far from done. Ratification in the U.S. will be very tricky; strong critiques from both left and right will hold it up until after the November election, and then it is anyone’s guess what happens. And the list of concerns arising in other countries is growing. (Don’t worry, though: TPP ratification in Brunei is still a slam dunk. What the sultan wants, the sultan gets!)

The TPP in its present form (including those damaging auto provisions) is not remotely set in stone. And by accepting the argument that Canada has no choice other than to go along with that flawed deal, or stay outside of it entirely, we are repeating Harper’s mistake of throwing all our cards on the table long before the game is truly finished. Our goal should be to try to fix those problems in the TPP. Resigning ourselves to playing only a take-it-or-leave-it role is too pessimistic. The argument about whether a flawed TPP is better than being outside of it entirely, is a complicated one, with many factors to consider; in no way is it self-evident that not joining would be the disaster  predicted by TPP advocates today. But at any rate, we don’t have to cross that bridge yet.

2. “Other studies have proven that free trade will help the auto industry.” I have spent a disproportionate amount of my adult life dissecting and critiquing quantitative free trade models (many of which use a bizarre and other-worldly methodology called computable general equilibrium modeling). In fact my Ph.D. dissertation was focused on this arcane subject, and sadly it seems this weird knowledge base is still as relevant today as it was back in grad school.

Remember that CGE models and other quantitative studies have been developed (in many cases supported by government funding) to show that Canada’s auto industry would be boosted by free trade with the U.S., Mexico, Europe, Korea, Japan, and now the TPP. In every case optimistic assumptions are made regarding the ability of factor markets to painlessly reallocate inputs and preserve employment, the automatic benefits of assumed productivity improvements for Canadian competitiveness and incomes, the balanced nature of trade flows, and more. These assumptions (not real-world empirical analysis) drive the always-sanguine results. The models predicted gains arise from many sources: including traditional comparative advantage factor reallocation efficiency gains (which tend to be quite small), and potentially larger effects arising from pro-competitive changes in industrial structure, scale, and the elimination of unobservable and unmeasurable frictions in trade. The more creative and extensive the assumptions about invisible frictions being eliminated, the bigger are the expected benefits of free trade. As I have shown in my critique of the CGE model which the Harper government commissioned to study the benefits of a CETA with the EU, those farther-reaching assumptions are equivalent to assuming the non-existence of the Atlantic Ocean! The resulting predictions of job-creation, income gains for every family, and vibrant balanced trade success bear no more relation to reality than does a SimCity game to our real lives.

The models never consider potential losses under free trade for countries which are not successful in maintaining or growing net exports after liberalization, and/or not successful in retaining and attracting mobile direct investment spending. Both of those problems have afflicted Canada mightily in recent years, which is why our trade performance since 2001 has been about the worst (measured by a number of indicators) of any industrial country. (And that miserable trade performance has been a key factor in Canada’s miserable employment and macroeconomic performance over the same period.) But don’t worry: so long as we assume full employment, balanced trade, capital immobility, incomes tied perfectly to productivity, and all the other features of Walrasian general equilibrium, we can assume away that painful real-world experience and get back to the happy world of eliminating Harberger triangles.

For detailed critiques of rose-coloured approaches to trade modeling, see my article: Economic Models and Economic Reality: North American Free Trade and the Predictions of Economists, International Journal of Political Economy 33(3), 2003, pp. 28-49. See also the detailed critique of trade models contained in my 2010 report on CETA for the CCPA (see especially Part 2), or the detailed critiques of CGE models of the Canada-Korea FTA that I prepared (together with Daniel Poon) for the CAW in the mid-2000s. Canada’s auto industry has been under pressure since the late 1990s, as soon as the industry began to adjust to the new policy context of the NAFTA. Expanded but highly unbalanced auto trade with Asia (now including free trade with Korea) and Europe has contributed to this decline, in addition to our enormous and growing auto trade deficit with Mexico. Given this consistently negative outcome, we should be all the more skeptical of the cheery forecasts of these simulation models.

The two reports cited by Mike are not CGE models, but share many of their assumptions. The Head and Meyer report is especially ambitious in its identification of assumed but unmeasurable additional frictions that are presumed to be eliminated by free trade (hence generating larger mutual benefits than traditional comparative advantage models). Specifically, this report confirms that the elimination of actual trade barriers (ie. measurable, observable tariffs) under TPP will be negative for Canada’s auto sector; in fact, Head and Mayer confirm that Canada experiences the largest negative impact of any TPP country. But the authors then assume two additional forms of trade frictions (related to technology transfer and marketing costs), which are also assumed to be eliminated under TPP, spurring pro-competitive changes in production costs and hence expanded output. In essence, the authors assume that the TPP will automatically improve productivity in Canadian auto plants (those run by Toyota and Honda in particular), and hence attract new business for them (including winning away business from Japanese and Mexican plants run by the same companies). This does not seem like a remotely credible hypothesis to me, and I suspect not to anyone else who works in the auto industry. I cannot remotely see how the TPP will make it concretely more efficient for Toyota and Honda to operate their Canadian plants (which are already 100% dependent on technology, engineering, and design imported by the companies from head office). The Toyota and Honda plants in Canada are already the most productive of any plants in North America. Production is assigned to auto plants not on the basis of marginal increments in productivity, but rather because of discrete all-or-nothing model allocations. If anything, the Toyota and Honda plants in Canada are the most in jeopardy under a TPP, because of how that deal would affect the decision by Japanese OEMs to produce here versus importing from Japan (all the more so given the TPP’s content rules, which mean that Japanese-branded vehicles could mostly be made in China yet still enter Canada tariff-free).

In summary, while the two reports cited by Mike are worth reading and make interesting contributions to this debate, in no way do they constitute evidence that my analysis is wrong.

3. “Minimum content and proportional effects.” The last set of counter-arguments advanced by Mike is that my assumption regarding the impact of weaker content rules on the sourcing decisions of North American auto producers are too pessimistic. First, he suggests, while the existing content rules set a minimum threshold, the industry actually operates above that threshold, and hence reducing the threshold may not affect ultimate sourcing decisions. Second, while the overall level of North American content in a typical North American-made vehicle might decline, Canada may not experience that loss proportionately; we may experience a smaller proportional loss than the U.S. or especially Mexico.

There is much room to debate these assumptions. I think that the realized average level of North American content in NAFTA-traded vehicles today (something around 75%) does indeed reflect the content rules (62.5% for vehicles, 60% for parts) negotiated when the NAFTA came into effect. Companies voluntarily go above that minimum for various reasons (including maintaining a safety margin, reducing transport costs and exchange rate risk, etc.), but the content rule sets a firm minimum to supply chain adjustments. Long-run decisions by companies about sourcing will very likely reflect changes in that minimum. My estimate does not require North American content levels to fall to the minimum only that they fall as much as the minimum was itself reduced (ie. by something under 20 percentage points). While we are not operating at the bare legal minimum today (and might not after a TPP, either), the fact that North American assemblers are allowed to offshore a majority of the content in their vehicles (and indeed must now compete with Japanese OEMs who have already done that, and would now have tariff-free access to North America) would undoubtedly lead to an offshoring of a significant incremental portion of total value-added.

And if companies are moving various parts of North American production offshore, will Canada experience a bigger, smaller, or just fair share of the resulting dislocation? I assumed the effects would be proportional, but I noted (and still believe) that that assumption is conservative. If anything, we may see a larger proportionate loss of production under the continental offshoring, because companies will likely shut higher-cost facilities in Canada and the U.S. first (retaining the newer, lower-cost facilities which have been built in Mexico in recent years). There are some factors which may mitigate against that result (for example, it may be that the sorts of parts made in Mexico are more sensitive to low-wage offshore competition than some of what is made in Canada although it is totally wrong to assume that Mexico only produces low tech  auto parts). I will stick with the proportionality assumption.

In conclusion, I do not think that any of the three concerns raised by Mike fundamentally weakens my argument that the TPP as currently written would eventually stimulate a significant offshoring of North American auto production (including to China and other countries which aren’t even in the TPP). There is very little conceivable upside to Canada’s auto industry in this deal (there will be no surge in Canadian auto exports to Japan, Malaysia, or Vietnam, for obvious reasons), and significant potential downsides (including the supply chain relocation discussed above, the growth of the existing imbalance with Japan, and others). The ultimate impact of those downsides could amount (over several years) to job losses measured in the tens of thousands. I will continue arguing for fundamental changes in the auto provisions of the TPP (and of course many other aspects of the deal, too like ISDS, patent laws, and copyright). And I think that it is quite possible politically to win those changes.