ANNEX
B
Arguments
Of The Parties
|
Contents
|
Page
|
|
Annex B-1
|
Executive
summary of the arguments of Canada
|
B-2
|
|
Annex B-2
|
Executive
summary of the arguments of Mexico
|
B-13
|
|
Annex B-3
|
Executive
summary of the arguments of the United States
|
B-26
|
ANNEX B-1
Executive
summary of the arguments of Canada
I. INTRODUCTION
1. This
arbitration is the culmination of six years of WTO litigation, beginning with
Canada's request for the establishment of the Original Panel in the WTO
proceedings concerning the United States' country-of-origin labelling
requirements. Canada prevailed at all
previous stages in this dispute. In this
proceeding, Canada is before the Arbitrator to defend its request for authorization
to suspend concessions to the United States in the amount of CDN $ 3.068
billion per annum, which reflects the devastating losses that continue to be
inflicted on the Canadian cattle and hog industries. The United States chose to request this
arbitration instead of bringing itself into compliance with the recommendations
and rulings of the DSB.
2. In
accordance with the DSU, and as demonstrated by previous arbitrations under DSU
Article 22.6, the United States must successfully challenge the accuracy of the
level of nullification or impairment proposed by Canada, rather than just
proposing alternative scenarios of its own.
Previous cases have also been clear that it is only after the United
States has met this burden, that the Arbitrator can assess alternatives to
determine the level of nullification or impairment. The United States has failed to make its prima facie case in discrediting Canada's methodology.
3. Canada and the United States agree
that the level of nullification or impairment should be calculated based on the
benefits that would accrue to Canada "but for" the amended COOL
measure: a comparison between the actual situation that existed following the
expiration of the reasonable period of time (RPT) to comply, and the situation
that would have prevailed in the period that follows the expiration of the RPT
if no WTO-inconsistent COOL requirements had ever been applied by the United
States. Critically, in measuring the level of nullification or impairment,
Canada's methodology takes into account any lingering effects of the COOL
labelling requirements.
4. Canada
drew attention at the meeting of the Parties to the real and devastating losses
experienced by Canada due to the introduction of the COOL requirements, by
highlighting the evidence in the marketplace that illustrates the damage. In particular, Canada focused on the fed-hog
industry, which suffered dramatic losses in export quantities. Fed hog exports
dropped from 48,778 per week to 10,051 per week after the introduction of the
COOL requirements. After the amendments
to the initial COOL measure came into effect in November 2013, exports fell
further to 7,833 per week through the end of 2014.
5. The
decline in fed-hog exports following the two-stage adoption of the COOL
requirements is so striking that restoring the volumes lost since September
2008 would require an increase of 500% from 2014 Canadian fed-hog export
levels. In addition to the huge decline
in export quantity, the fed-hog industry also suffered a loss of price. Comparing the pre-COOL period to the period
after the amended COOL measure, the fed- hog price basis fell by about $0.09
per pound or about 11% based on 2013/14 prices.
6. These
remarkable losses that occurred in the real hog market should be compared to
the simulated world imagined by the United States, which suggests a negligible
drop in imports and price compared to reality. The drop in exports following
the introduction of the COOL requirements was large for each of the four
livestock categories. In contrast, in
every case the U.S. model suggests that the impacts of the COOL requirements
were tiny.
7. Canada's
approach remains superior because it is a methodology that accurately captures
the impact of the amended COOL measure on the Canadian cattle and hog
industries. It does so by using
real-world data, and by accurately capturing the differential impact of the
amended COOL measure on the Canadian cattle and hog industries. The model proposed by the United States fails
to do this as it relies on simulations without any connection to reality, and
does not use real data to measure the impact of the amended COOL measure.
II. Canada's Methodology is the only Appropriate Methodology to Measure
Canada's Losses
8. Canada's
methodology is the only appropriate method of assessing losses in this case
because it accurately and directly measures the level of nullification or
impairment suffered by Canada. Canada
uses an econometric analysis to assess the losses, which employs real-world and
appropriate data that capture the differential impacts of the COOL requirements
on Canada's cattle and hog industries.
9. Econometric
modeling is appropriate in this case because such models are routinely used to
estimate the market impact of a regulatory event. The assertion by the United States that an
econometric approach should not be used in a DSU Article 22.6 arbitration is
unreasonable and contradicts the basic economic principle that real-world data
should be used if they are available.
Moreover, the United States argument that it should not be used in this
context is untenable, given that the U.S. model relies on elasticities
generated by econometrics.
10. The
econometric model submitted by Canada is also the most direct way to assess the
impact of the COOL requirements on Canada's cattle and hog industries. The purpose of such a model is not to capture
all the complexities in the market, and reflects the current practice among
leading economists to reduce complexity and focus carefully and parsimoniously
on the specific policy question at hand.
11. At
the heart of Canada's analysis is the differential impact of the amended COOL
measure on the price of Canadian livestock and the quantities exported. Canada's Methodology Paper captures this
differential impact, which is reflected in a widening of the price basis
between fed and feeder cattle and hogs in the United States compared to those
exported from Canada, as well as a reduction in import quantities of Canadian
livestock. Canada's model is the
appropriate model to use in this proceeding because the U.S. model, as
explained below, fails to incorporate the differential impact of the amended
COOL measure and therefore does not accurately capture the losses suffered by
Canada.
12. Further,
Canada's model focuses solely on the impacts of the labelling requirements and
avoids including extraneous variables that would introduce bias to the measured
impacts of the amended COOL measure.
More specifically, Canada only included variables that have a
differential impact on either the price basis or quantity. The inclusion of any others is not needed for
Canada's approach of isolating the impact of the amended COOL measure.
A. Price Basis Captures the Differential Impact
of the COOL Requirements
13. Canada's
methodology uses the price basis as the dependent variable as it allows for the
isolation of the differential effect of the COOL requirements on the export
price of Canadian cattle and hogs destined for the U.S. market. It does this by automatically capturing the
impacts of a host of variables that affect livestock prices in both countries
in a similar way. With these factors
controlled for in the price basis, the methodology need only include those
variables that represent factors with differential impacts on Canadian and U.S.
livestock prices. Canada has done so, by including variables for the exchange
rate, seasonality, two BSE events (for cattle), and the closing of the Maple
Leaf Foods plant (for fed hogs). These
differential variables cause fluctuations in the price basis, and they have
been included on the right hand side of the equations to ensure that they are
controlled for in Canada's methodology.
14. To
use the absolute price in Canada's methodology, as the United States advocates,
would mean that the effects of numerous other variables that affect U.S. and
Canadian livestock demand and supply would not be effectively controlled for in
the methodology. As a result, any rise
in the Canadian price that is caused by other factors (including those that are
inherently controlled for in the price basis), could be erroneously interpreted
as being caused by the amended COOL measure.
Further, to use the U.S. price as an explanatory variable would create
bias in the regression model because it is endogenous. This means the U.S. price is correlated with
the error term in the regression model and all estimated coefficients in such
an equation are biased. Because many
additional variables would need to be included to control for COOL, statistical
issues would arise leading to the problem of spurious regressions. Despite these major problems, in response to a
request by the Arbitrator, the United States attempts to use absolute price in
Canada's methodology. The result is an
internally incoherent model with irrational results. In fact, the results
actually favour Canada as the effect of the COOL requirements become cumulative
in such a methodology.
15. The
United States asserts that price basis exaggerates the estimate of the level of
nullification or impairment because it includes any increase in U.S. price that
may be a result of the COOL requirements.
The United States is incorrect because the non-discriminatory shared
costs of COOL that increase the U.S. price also increase the Canadian price;
likewise, any non-discriminatory influence of COOL that causes a decrease in
the U.S. price would decrease the Canadian price by the same amount. These
effects are not reflected in the price basis.
Moreover, any impact of the COOL requirements on raising the U.S. price
is minimal because Canadian imports of cattle and hogs only represent between
2% and 4% of the massive U.S. market. To
suggest that the COOL requirements would raise the U.S. price when the Canadian
share of the market is so tiny just does not make any sense. Indeed, this
conclusion is reflected in the Tonsor et al. (2015) report commissioned by the
USDA.
16. The
United States has incorrectly characterized the discussion in Pouliot and
Sumner's Food Policy article (Exhibit US-35
(22.6)) as supporting its position that price basis should not be used in
Canada's methodology. However, the
authors of that paper did use price
basis for their analysis because it was and is the most appropriate measure to
determine the impact of the COOL requirements on price. Further, the United States has tried to argue
that Canada's approach is at odds with the academic literature generally, but
in its own submission it cites more articles that use the price basis rather
than absolute price (see U.S. comments on Question 35).
B. The Variables Proposed by the United States
Do Not Meet the Criteria for Inclusion
17. Canada
has consistently argued that the explanatory variables proposed by the United
States do not belong in Canada's methodology because they do not meet the
criteria for inclusion.
18. Canada
has shown that the variables proposed either do not have a compelling economic
rationale as to their differential impact, have not been supported by empirical
evidence by the United States, or both.
Canada specifically addressed variables proposed by the United States in
respect of (i) economic fluctuations or recession; (ii) feed costs; (iii)
transport costs; (iv) other competing imports; and (v) drought. Each of these is addressed summarily below.
1. Economic Fluctuations or Recession
19. There
is no compelling economic reason for inclusion of this variable as although unemployment
in the United States may have led to less meat consumption, this change did not
differentially affect demand for livestock buyers for U.S. animals as compared
to Canadian animals. The United States
has not been able to provide any empirical evidence to suggest otherwise.
20. Further,
the United States made a serious specification error in using a variable that
reflects only the recession in the United States, when it is the differential
impact of the recession that would be relevant.
Canada has also noted that any attempt to measure the differences in the
two recessions would involve imprecision due to time lags in macro-economic
conditions and the impact on export shipments or price basis. Even the best proxy data available would have
at most a small and indirect influence on export quantities, and none at all on
price basis.
2. Feed Costs
21. Potential
feed cost variables that affect export quantity or price basis would need to
reflect any movement in the difference of costs per unit of weight gain of
cattle and hogs between Canada and the United States, as it is the relative
cost of a pound of weight gain that affects the location of where an animal is
raised in its lifecycle. Appropriate
econometric specification would require developing some information and
assessment of livestock producer expectations of future costs of feed and feed
rations, and Canada knows of no such data.
Further, the United States has not provided any empirical evidence that
such a variable would have a differential impact on prices or quantities. In any event, even if there are fluctuations
in feed costs, these would not cause a U.S. livestock buyer to pay more or less
for a U.S.-born animal as compared to a Canadian-born animal.
3. Transport Costs
22. In
theory, transport costs could affect the price basis and export
quantities. However, the United States
has provided no empirical evidence that transport costs have affected price
differences between Canada and the United States. In fact, the United States has provided
evidence with respect to fuel costs in the United States only. Canada provided evidence on transportation
costs with respect to feeder-pig transactions covered by the Procedures of the Arbitrator on Business Confidential Information,
which cannot be summarized here.
4. Other Competing Imports
23. Imports
from Mexico could in theory affect the quantity of imports from Canada, but
there is no empirical evidence that they have done so. Indeed, such an impact is highly unlikely
given the geographical separation of Canadian and Mexican imports, and the size
differences in the feeder cattle imported from the two countries. With respect to price, any influence of
imports in the market would affect prices of comparable U.S. and Canadian
animals to the same degree, and would be very small in any case because Mexican
feeder cattle comprise a trivial share of the huge U.S. feeder-cattle
market. Further, because this variable
is endogenous, controlling for Mexican imports by including the variable in a
price basis estimating equation would yield biased estimates.
5. Drought
24. The
empirical evidence suggests that the theory put forward by the United States,
that the impact of the drought in the United States (and Mexico) would have led
to an increase in supply from Mexico and a decrease in demand for Canadian
feeder animals, is incorrect: feeder-cattle exports from Canada increased due
to the drought by well over 100,000 head from 2013 to 2014. Regardless, the feeder cattle in Canada are
not substitutable for those in Mexico.
Further, the United States has not provided any evidence to suggest that
the drought affected the price of Canadian feeder animals compared to U.S.-
origin feeder cattle. Designing a
variable to take into account the drought would be difficult because of the
complex timing of events, the cattle cycle and producer expectations.
C. The Econometric Estimations with the United
States Proposed Variables
25. At
the request of the Arbitrator, Canada has produced estimations with all of the
above variables included in its estimations, both individually and
cumulatively, to determine their effect on the price basis. The regressions for all of the above
variables are included in Exhibit CAN-68 (22.6) through Exhibit CAN-76
(22.6). Overall substantive and
statistical results changed little with the addition of these variables. Some
of the DCOOL1 and DCOOL2 effects are higher, and some are lower. As the United States has admitted, there is
no systematic pattern of changes compared with the results in Canada's
Methodology Paper. Therefore, there is
no error in Canada's approach. Further,
these outcomes align with the results of the estimations done at the request of
the Original Panel.
26. As
when included separately, most estimated coefficients of these variables
included all together are not statistically significant. Nevertheless, the inclusion of all the
variables would increase Canada's export losses by $210 million, to a total of
$2.254 billion. Adjusting for the errors
the Arbitrator found in its Table related to Question 42, Canada's total losses
increase even more, from $2.981 billion to $3.234 billion. Even when the addition of these variables
results in higher losses to Canada, Canada's position continues to be that the
losses calculated in its Methodology Paper remain more appropriate and
accurate.
27. Contrary
to the requests of the Arbitrator, the United States did not produce
estimations with all of the above variables in all of its price calculations
individually, and cumulatively. Instead,
the United States limited its analysis to 850-pound feeder cattle only,
forgoing any estimations on the other weight classes of feeder cattle, and the
other three categories of livestock entirely.
Not only has the United States failed to provide any empirical evidence
for its position that these variables should be included, it has also
cherry-picked one category, which is the weight class for which the price basis
effects of COOL are the smallest.
28. Regardless,
even when the United States uses its mis-specified variables, the estimations
that result favour Canada or are neutral to Canada. More specifically, when the United States
includes the recession variable the effect of COOL on the price basis is significantly
larger than in Canada's Methodology Paper.
However, despite this result that favours Canada, Canada continues to
take the position that this variable should not be included. For both feed costs and transportation costs,
the inclusion of these variables by the United States changes the results very
little. When Canada corrects for the
error the United States makes in its estimations for monthly feeder-cattle
exports, the results are similar to those in Canada's Methodology Paper. Canada was unable to replicate the data the
United States used for the drought because the United States failed to provide
interpretive information. In any case,
the U.S. estimates only have a slight impact on the price basis.
29. When
these variables are added by both the United States and Canada, the results
change little. This proves that Canada's
model is robust and its specifications are accurate.
D. Canada uses Real-World and Accurate Data to
Assess its Losses while the United States does not
30. An
econometric approach requires real-world marketplace data on the situation that
prevailed before and after the measure at issue. In the current case, detailed, official
government and industry data do exist before and after the amended COOL
measure. In these circumstances, it
makes no economic or empirical sense to ignore the wealth of data
available. This is especially true when
the alternative U.S. model is based solely on theory, and simulations built on
faulty assumptions. It is not only
possible, but highly preferable to use the available marketplace data in any
calculation of Canada's losses.
31. Canada's
Methodology Paper outlines the data sources it uses in its econometric
assessment of its losses in Appendices I and II. The weekly data go back twelve years in the
case of hogs, and ten years for cattle.
They are drawn from both U.S. and Canadian government sources (with the
exception of feeder pigs where no such data are available), as well as
industry, and are the most detailed, accurate, relevant and specific data for
the task at hand. The United States
takes issue with Canada's choice of the U.S. government data set, as if this
were a major flaw with Canada's methodology.
This is simply not a valid argument. U.S. Department of Agriculture
APHIS data are derived from actual border inspections of livestock shipments,
as opposed to the values of shipments and associated quantities. The USDA APHIS data are valid, accurate and
detailed. Further, the Original Panel considered
the use of USDA APHIS data as being acceptable and reliable.
32. In contrast,
the United States uses no real-world data to determine the impact of the
amended COOL measure or to determine how the markets have responded. The only real data the United States uses is
for its 2014 base-year calculations. The
United States has essentially abandoned the wealth of readily available data
and adopted a model that is suitable for use only when there are no such data
available.
33. The
United States also did not use appropriate data for monthly imports of feeder
cattle when requested to by the Arbitrator.
It not only failed to use the data it suggested it would (U.S.
Department of Commerce data), but then used an inaccurate USDA summary data set
that includes only some of the relevant livestock categories, leading to
misleading and incomplete results. The United States also mistakenly specified
the key COOL variables. When these
errors were corrected by Canada, the results are very similar to when weekly
data are used. None of the effects is different to a statistically significant
degree and all effects are strongly and significantly negative.
34. Further,
at the request of the Arbitrator, Canada used monthly data instead of weekly
data in its estimations. The results are
found in a table in Canada's responses to Question 37. Despite some expected
differences, the results are very similar to the results using weekly
data. None of the differences is
statistically significant. In fact,
using monthly data causes the estimate of Canada's losses to rise above the
$3.068 billion that Canada reported in its Methodology paper. Also at the
request of the Arbitrator, Canada changed the base period by several weeks,
extended the sample period for cattle, changed the dates of the COOL dummies,
and used a different implementation date for fed cattle, larger feeder cattle
and fed hogs. The results were generally
not significantly different from those in Canada's Methodology Paper and any
changes were predictable.
E. Canada's Data for Feeder Pigs are Accurate
and Verifiable
35. The
parties agree that there are no available feeder price data amenable for
statistical analysis in this case. As a
result, Canada provided evidence covered by the Procedures
of the Arbitrator on Business Confidential Information, which cannot
be summarized here. In any event, Canada's
evidence is further supported by econometric analyses and simulations. It is
also consistent with the losses suffered in the three other livestock
categories.
III. Methodology
and calculation of price suppression losses
36. In
addition to the loss of revenue resulting from the reduction in livestock
exports to the United States and the lower prices received for those livestock
that were exported, the amended COOL measure has caused Canadian livestock
producers to suffer losses resulting from the reduction in the price received
in Canada for cattle or hogs that were not shipped to the United States.
37. Canada's
estimate of these losses is carefully constructed and conservative. First, it avoids double counting of
suppressed exports that are already included in the estimate of the loss of
export revenue. It does this by ensuring
that in the price-suppression equation, the quantity of animals in Canada to
which price suppression applies is reduced by the implied loss of quantity of
exports that are estimated to be part of the loss of export revenue. Second, it does not include the reduced
production and supply from Canadian livestock industries due to the lower
prices that they face.
38. The
losses due to domestic price suppression for all categories of livestock except
feeder pigs were based on appropriate industry and government data. The domestic losses for feeder pigs were
calculated on the basis of the most reliable data Canada could access. Both parties
agree that there is no public government source for data that provides
consistent time-series price data amenable for statistical analysis of the
price of feeder pigs in Canada.
Therefore, Canada relies on evidence covered by the Procedures
of the Arbitrator on Business Confidential Information, which cannot
be summarized here.
39. A
high level of integration between the Canadian and U.S. markets results in
price arbitrage between the Canadian export price of live cattle and hogs and
the price of domestic cattle and hogs.
As a result of the amended COOL measure, once a drop occurred in the
export price, the price of the domestic livestock adjusted accordingly. This arbitrage mechanism is well accepted as
a "given" in the industry.
There are no domestic factors that impede this mechanism. Canada is a price taker for all categories of
cattle and hogs as it supplies only a trivial market share for each in the
massive U.S. market (in all cases less than 4%, and in most cases less than
2%.) There is therefore a direct causal
link between the amended COOL measure and the domestic price-suppression losses
experienced in Canada. Because of this
arbitrage between markets, the price impacts for the domestic price-suppression
analysis are the same as those that are used in the calculation of export
revenue losses.
40. In
response to Question 42 from the Arbitrator, Canada acknowledged an inadvertent
error in its original calculation of price-suppression losses for feeder pigs
(of $325,400,000, as compared to the Arbitrator's calculation of the same losses
in the amount of $237,805,000). As to
the price-suppression losses claimed by Canada in respect of the other three
animal categories, it explained the small discrepancies between the numbers
contained in Canada's Methodology Paper and those contained in the Tables to
Question 42 from the Arbitrator.
IV. Losses Due to Domestic Price Suppression Are Supported by both the
DSU and the Jurisprudence
41. The
amended COOL measure impaired Canada's benefit by adversely affecting the
conditions of competition of Canadian livestock exports to the United States,
which resulted in reduced prices and export volumes. As a direct result of the violation of
national treatment and the highly integrated nature of the two markets, the
livestock that were not exported to the United States also received a lower
price in Canada. The resulting increased
supply in Canada and the lack of an alternative export market suppressed the
prices of these animals in the Canadian market, resulting in specific and
quantifiable losses.
42. Based
on the highly integrated nature of these two markets, these are direct losses
from the denial of a direct benefit. In
the alternative, domestic price suppression losses are at the very least losses
that result from the impairment of an indirect benefit of national
treatment. Regardless, both direct and
indirect benefits are covered by DSU Article 3.3 and therefore by DSU Article
22.4.
43. The
United States concedes that the nullification or impairment suffered by Canada
in this case includes export losses.
However, the United States challenges Canada's claim for domestic
price-suppression losses, arguing that these losses do not flow from one of the
"benefits" accruing to a Member under the provisions of the covered
agreements, in this case the GATT 1994 and the TBT Agreement. The United States
is incorrect.
44. In
this proceeding, the denial of the "benefit" in question is based on
market access, as the United States acknowledged at the meeting of the
Arbitrator with the parties. As Canada
has explained, its entitlement to national treatment under the GATT 1994 and
the TBT Agreement provides a benefit not to see the competitive positions of
Canadian cattle and hogs adversely affected by the amended COOL measure,
including through domestic price effects.
The jurisprudence supports this position. As was stated in US-Byrd
Amendment, the denial of benefits accruing (which is a broader
concept than the breach of an obligation) is not to be confused with the
violation of the obligations itself.
45. Contrary
to what the United States argues, there is no indication that losses need to be
limited to those suffered in the territory of the respondent. Indeed, the
indirect losses claimed by requesting parties in US-Byrd
Amendment were in the form of exports to other markets. These were not rejected in principle, but
because they were too speculative and unquantifiable. A limit to the scope for a claim of indirect
losses was decided in EC-Bananas
where the panel rejected a losses claim based on exports through a third
country, because the claim was based on inputs into the final product made by
another country. In contrast, Canada is
the direct exporting party, the price-suppression losses are in its own market.
46. Further
and crucially, indirect losses, like direct losses, require causation by the
impugned measure, and verifiable information to support the claim. Canada's claim has both. Causation occurs through the arbitrage mechanism,
discussed above, and Canada's careful calculation of these losses is based on
verifiable government and industry data.
47. The
United States has repeatedly failed to define the "benefits" to which
it refers; instead, it vaguely states that these benefits "relate to trade
effects". It erroneously relies on cases interpreting the level of
nullification or impairment as being concerned with the "trade
effects" of the impugned measure, in order to claim that the jurisprudence
has limited the scope of "trade effects" to export losses. This is a mischaracterization of the jurisprudence. On the contrary, WTO jurisprudence suggests a
broad interpretation of "trade effects". Further, there is nothing in the DSU that
limits the level of nullification or impairment to export losses.
48. Contrary
to what the United States argues, cases that focus on the "trade
effects" of nullification or impairment are not limited to an assessment
of losses based on exports alone. For
example, in US-1916, "trade effects" included the settlement awards and
judgments of the 1916 Anti-Dumping Act against the complainant's companies, and
neither involved an assessment of trade flows.
In US-Byrd
Amendment, the Arbitrator rejected a narrow
interpretation of "trade effects", dismissing the U.S. position that
nullification or impairment was limited to the direct trade loss resulting from
the violation.
49. It
is incorrect for the United States to argue that all domestic losses claims are
inadmissible or ultra
vires the DSU.
In US-Gambling,
a claim for domestic losses was based on the
"multiplier effect" of the impugned measure on the broader Antiguan
economy. The Arbitrator did not reject
the claim for domestic losses because it was incompatible with the DSU but
because the claim contradicted some of the other Antiguan arguments.
50. There
are no WTO arbitrations under DSU Article 22.6 that have rejected a claim for
domestic price-suppression losses. In
the two cases that the United States cites as support for its argument that
"trade effects" are limited to export trade (EC-Hormones
and EC-Bananas), the complainants only
claimed and argued for export losses.
51. Unlike
the cases of EC–Hormones, EC–Bananas,
or other "trade effects" cases, in which the markets concerned are
not integrated, the live cattle and hog markets of Canada and the United States
were almost fully integrated before the adoption of the original COOL
measure. The live cattle and hog
industries in both countries are structurally similar and interdependent. As a result and as the United States has
recognized, any change in supply or demand in the United States will directly
and causally affect Canada.
52. Further,
there is absolutely no merit in the argument made by the United States that, if
the level of suspension of nullification or impairment includes domestic
price-suppression losses, the level of suspension has to be decreased by a
calculation of the "broader economic effects on the U.S.
economy". The level of
nullification or impairment is an independent calculation based on the losses
of the complaining party. There is no
calculation with respect to the level of suspension of concessions. The meaning of "equivalence" is
that the level of suspension of concessions will equal the level of
nullification or impairment that is determined by this arbitration.
53. Finally,
the burden is on the United States to make a prima facie
case that the level of suspension is not equivalent to the level of
nullification or impairment. In
comparing the precise domestic price suppression losses claimed by Canada, to
the speculative impact that the suspension could have on the U.S. economy, the
United States has failed to meet its burden of proof.
V. The EDM is Misspecified and Inappropriate in these Circumstances
54. As
explained above, the United States has failed to meet its burden of proof
because it has not demonstrated that Canada's proposed level of suspension of
concessions exceeds the level of nullification or impairment that resulted from
the COOL labeling requirements.
55. Instead
of fulfilling this burden, the United States has presented a model that is
conceptually ill-suited for this case.
However, more importantly, the Equilibrium Displacement Model (the
"EDM") proposed by the United States utilizes incorrect elasticities,
fails to consider differential segregation and compliance costs, fails to use
real-world data outside of the base year, and is rife with faulty
assumptions. The model also suffers from
serious numerical errors. As a result,
the U.S. model is fundamentally flawed and severely underestimates the losses
faced by Canada.
A. The EDM is Inappropriate in these
Circumstances
56. Economists
are clear that when appropriate market data are available, an econometric
analysis is preferable to a simulation model.
Indeed, hundreds of refereed articles in the economics literature have
examined data before and after the introduction of a government measure by
using econometrics, including in the most up-to-date peer-reviewed academic
literature. There is simply no reason
not to adopt an econometric methodology in this case. Nevertheless, the United States has adopted
an EDM, which is a simulation model that does not use real-world data on the
actual impacts of the amended COOL measure.
57. A
simulation analysis is best suited for assessing the future impact of a policy
change or measure, when the impact of a policy or measure has not yet been
observed. This is precisely why a
simulation analysis was used in Brester et al., Lusk et al., and Tonsor et al.
These studies were done before the policy was put in place and thus are
predictive only, and do not benefit from real-world data or the knowledge of
how the COOL measure was actually applied.
B. The EDM Uses the Wrong Elasticities
58. The
EDM uses short-run elasticities when long-run elasticities should have been
used. Short-run elasticities are appropriate in situations in which very little
time is needed for the domestic market to reach equilibrium. In this case, the Canadian market has still
not completely adjusted after six years, and therefore the use of short-run
elasticities is inappropriate. Further,
using short-run elasticities is inconsistent with the underlying assumption in
the EDM of adjustment from one full market equilibrium to another with no
adequate market adjustment time. The United States has in fact admitted that it
may not have chosen the correct elasticities for this case.
59. As
an example, the United States' model relies on a study in Brester et al., which
obtained low export-supply elasticities using Canadian export-share data that
are no longer applicable and by using short-run elasticities over an adjustment
period of just three months. In reality,
cattle-supply adjustment takes much longer than this. The use of long-run elasticities and the
current export-share data would have produced much higher export supply
elasticities, and as a result, much higher losses. This assumption of full equilibrium using
short-run elasticities highlights an internal inconsistency in the EDM that
leads to a major underestimation of losses suffered by Canada.
60. The
United States has also erred in using inappropriate "import supply
elasticities" (which should be called export supply elasticities) for U.S.
imports of feeder and slaughter animals, which are equivalent to Canadian
exports of these animals. In essence,
since no such elasticities exist, the United States sets these categories to
the supply elasticity for U.S. imports of wholesale meat. Similarly, the U.S. assigns a certain
elasticity value to beef import supply and two livestock import-supply
elasticities, with no explanation. These
arbitrary assignments further undermine the U.S. model.
61. The
crucial point is that the United States has used export supply elasticities
that are far too small for each category of animals from Canada and feeder
cattle from Mexico. This is a major
error because export supply elasticities are one of the most important
parameters in a simulation model since they determine the magnitude of the
different effects on export quantities and prices in a simulation model.
Specifically, the United States has used the wrong underlying domestic
(Canadian and Mexican) elasticities and market shares in the export supply
elasticity formula, which resulted in artificially low figures. To illustrate this grave flaw in the EDM,
Canada provided a chart containing the proper elasticities in its Comments on
the Responses of the United States to Question 46 of the Arbitrator (dated 8
October 2015). The United States
position that the elasticities proposed by Canada are not supported by academic
literature is completely false, and in any event, they are determined by a
standard formula included in the literature as well as in Canada's submissions.
62. A
stark example of the U.S.'s flawed export supply elasticities is found in the
responses to Question 46 of the Arbitrator.
Canada has shown that it was the larger export shares of Canadian fed
hogs and feeder pigs sourced from the dated Wohlgenant study that resulted in
much smaller implied export-supply elasticities than those obtained by Canada
using current export shares. Even worse,
for feeder cattle the United States failed to download the proper export data
from their own source table and use clearly misrepresented elasticities from
their own source document. Canada uses the much more recent supply and demand
elasticities from the Tonsor et al. (2015) report, also cited by the United
States, to illustrate the correct calculation of export supply elasticities
with the proper export share. Similarly,
in response to Question 51, Canada also showed that it is possible to obtain
proper elasticities, which account for market realities, when appropriate data
are used.
C. The EDM Fails to Account for Differential
Segregation Costs
63. One
of the most fundamental errors in the EDM is its failure to take into account
differential compliance and segregation costs.
A proper simulation analysis would consider how market forces determine
how imports of Canadian cattle and hogs are traded relative to the equivalent
U.S.-origin animals. Indeed, this is the
very rationale for Canada's complaint. The COOL labeling requirements have
undeniably resulted in greater implementation costs for U.S. firms that use
imported livestock as opposed to U.S. firms that exclusively use
domestic-origin livestock. However, in the EDM, the U.S. domestic supply chain
is assumed to have the same segregation costs as a supply chain that uses both
imported and U.S.-origin livestock. This
assumption has fatally biased the results of the EDM. For example, Equations 18-23 of the U.S. EDM
contain the misplaced U.S. assumption that changes in the price of U.S.
livestock must equal changes in the price of imported livestock of the same
weight categories.
64. The
United States' denial of the existence of differential compliance and
segregation costs has been clear throughout these proceedings. While the United States continues to claim
that the EDM reflects the reality of the U.S. livestock and retail beef and
pork markets, this is totally inaccurate.
As both Canada and Mexico have noted, the EDM fails to consider the
crucial fact that the COOL measure influenced buyers to treat livestock of
different origins differently in order to comply with the labelling
requirements. Accordingly, the
inaccuracy of the EDM could only be corrected through a re-structuring that
accounts for these differential costs imposed on all intermediate buyers of
livestock. However, the United States
refused to do this, even when requested to do so by the Arbitrator in Question
44.
D. The EDM is Plagued by Additional Faulty Underlying
Assumptions
65. In
addition to the most serious flaws explained above, the results of the EDM are
biased by a host of other faulty assumptions, as Canada outlined in its
submission of 12 August 2015.
66. For
example, the EDM assumes perfect competition amongst firms in the U.S. meat
processing industry such that individual firms do not influence prices. The EDM also assumes efficiency of all plants
regardless of size, that all animals are used in all markets by all plants and
firms, and that all markets clear in a complete and perfect equilibrium both
with and without the amended COOL measure.
Furthermore, it assumes that there are no market imperfections of any
sort, and no path for adjustment or transition.
67. These
assumptions are obviously at odds with reality and therefore provide a flawed
foundation for the EDM. Because the EDM
is based on these assumptions as well as others listed above, its results are
completely unreliable.
E. Reliability of the Data and Numerical Errors
68. Canada
finally notes the many numerical errors that were made in the EDM calculations,
as discussed in Comments on the Responses of the United States to the Questions
of the Arbitrator (dated October 8 2015).
Moreover, the reliability of the data used in the EDM calculations is
seriously flawed because some studies relied upon are decades old and
unpublished.
69. Further,
the United States has committed a series of serious data errors in its
submissions. For example, it uses United
States Census data for the base year even when they are not suitable because
they do not provide information specific to imports of barrows and gilts for
slaughter. The United States arbitrarily
divided total hog imports by half to create the data for the fed hog category.
70. In
short, a simulation cannot replace the need to examine the impact of the
measure in the real world. As demonstrated in Canada's calculations using the
EDM in response to Question 51, accurately calculated elasticities that account
for market realities can yield results that are plausible. However the U.S.
calculations in the EDM are based on inaccurate and unrealistic elasticities.
These theoretical and practical flaws in the EDM illustrate the
inappropriateness of using a simulation model in these circumstances and
further demonstrate the suitability of Canada's econometric approach.
VI. CONCLUSION
71. The United States has failed to make its prima facie case to discredit Canada's methodology to
determine the level of the nullification or impairment of benefits it is
entitled to. Canada has demonstrated
that its methodology is sound and accurately represents the devastating losses
Canada has been suffering for seven years.
72. Indeed, when Canada includes the variables
proposed by the United States, uses monthly instead of weekly data, and changes
dates as requested by the Arbitrator, the results do not change significantly,
which demonstrates that Canada's methodology is robust. Similarly, even when the United States makes the
changes to Canada's methodology requested by the Arbitrator, the results do not
alter significantly and do not suggest that Canada's approach is flawed. The
fundamental integrity of Canada's Methodology Paper remains intact and its
calculations are accurate. Therefore,
the Arbitrator should use Canada's methodology to award losses suffered by
Canada in this case.
73. Canada respectfully requests that the
Arbitrator find that Canada's level of nullification or impairment is as was
specified in Canada's request to the DSB for authorization to suspend
concessions to the United States, i.e. CDN $ 3.068 billion per annum, subject
to a correction conceded by Canada in paragraph 40 of this Executive Summary.
ANNEX
B-2
Executive
summary of the arguments of Mexico
1. Mexico's Methodology Paper demonstrates that the COOL measure has
reduced the prices paid for Mexican cattle in the United States, reduced
Mexican export sales and supressed the prices of feeder cattle in the Mexican
domestic market. Because the United States has not taken any steps to bring the
COOL measure into conformity with its obligations, Mexico is seeking
authorization to suspend concessions in the amount of USD 713.4 million.
2. In the present case, the Arbitrator's mandate under Articles 22.6
and 22.7 of the DSU is to determine whether the proposed level of the
suspension of concessions requested by Mexico is equivalent to the level of the
nullification or impairment of benefits accruing to Mexico as a result of the
United States' failure to bring its WTO-inconsistent COOL measure into
compliance. The level of the nullification or impairment that Mexico has
suffered is the difference between the actual
level of benefits accruing to Mexico at the time of expiry of the reasonable
period of time (RPT) to comply as a result of the adverse effects of the COOL
measure and the level of benefits that would have accrued to Mexico in a
counterfactual scenario in which the COOL measure had never been adopted. This
approach properly measures the full extent of the nullification or impairment
caused by the COOL measure because it measures its adverse effect against the
level of benefits that the parties negotiated under the covered agreements —
i.e., benefits that prohibit the COOL measure from ever existing.
3. Mexico's methodology uses an approach that combines detailed econometric estimates and
simulations to accurately calculate the level of nullification and impairment.
First, an econometric model using observed data is employed to estimate the
adverse effects of the COOL measure on the price of Mexican feeder cattle
exported to the United States. Second, this estimated export price impact is
then used to simulate a consistent reduction in exports of Mexican feeder
cattle. Third, the estimated export price impact is used to measure the corresponding
price impact in Mexico's domestic market. Export losses and price suppression
losses are then calculated based on the adverse effects of the COOL measure.
4. In
response, the United States proposes a novel simulation that builds on the currently observed market
equilibrium and removes the costs associated with the COOL measure. The approach used by the United States is
flawed in many ways. First, instead of using the available pre- and post-COOL
data to directly estimate the adverse effects of the COOL measure, the United
States purports to measure the impact of removing the COOL measure through a
complicated theoretical model that is inconsistent with the actual effect of
the COOL measure as found in the WTO rulings. Second, the United States' concept of
calculating losses based on market outcomes with the COOL measure in place is fundamentally flawed as it assumes, inter alia,
that the impact of imposing a measure is equivalent to that of removing
it. Importantly, under the approach
proposed by the United States, if the harms caused by a measure are not fully
reversible, then the amount of nullification or impairment would inevitably be
understated.
5. In US – Upland Cotton (Article 22.6 – US),
the Arbitrator confirmed that the party objecting to the proposed
countermeasures bears the initial burden of establishing a prima facie
case that the proposed countermeasures are not in accordance with the
requirements of the relevant WTO provisions; if this initial burden is
discharged, then it falls to the party proposing the countermeasures to rebut
such a presumption.[5] This standard is consistent with the burden of proof applied in other
arbitrations, including EC – Hormones (US)
(Article 22.6 – EC), EC – Bananas III (Ecuador)
(Article 22.6 – EC) and US – 1916 Act (EC)
(Article 22.6 – US). Similarly, the arbitrator in US – Gambling (Article 22.6 – US) confirmed that the burden
rested upon the United States, as the objecting party, to demonstrate that the
level of suspension proposed by Antigua was not equivalent to the level of
nullification or impairment resulting from the continued application of the
WTO-inconsistent measure. For the purpose of discharging this burden, the
Arbitrator emphasized that the United States must successfully challenge the
accuracy of the level of the nullification or impairment reflected in the
counterfactual scenario proposed by Antigua, rather than merely proposing
alternative scenarios of its own.[6]
6. Thus, the United States bears the initial burden of establishing a prima facie case that the level of suspension of benefits
requested by Mexico is not in accordance with the requirements of the DSU. The
United States has failed to discharge its burden. There are significant legal
and conceptual errors in the United States' criticisms of Mexico's methodology
and also in the alternative methodology that the United States proposes. Mexico's
comprehensive analysis is the correct approach under the circumstances, and it
has been properly applied to accurately estimate the level of nullification and
impairment caused by the amended COOL measure. The United States' inaccurate
criticisms and flawed alternative "Equilibrium Displacement Model"
(EDM) are therefore insufficient to establish a prima facie
case the Mexico's methodology is inconsistent with DSU Article 22.4.
7. In the event that the Arbitrator disagrees and finds instead that the
United States has established a prima facie
case, in whole or in part, then Mexico submits that its methodology remains the
most appropriate approach to assessing the level of the nullification or
impairment of the benefits that would accrue to Mexico but for the adverse
effects of the amended COOL measure, subject to any adjustments that the
Arbitrator determines are required.
8. Mexico emphasizes that, although Canada's approach is very similar
to Mexico's, the Canadian methodology is not identical to that of Mexico. In particular, Mexico is able to rely on
pricing data for Mexican cattle from sales within the United States, and that
eliminates the relevance of certain variables discussed by the United States.
For this reason, Mexico's analysis must be reviewed independently and evaluated
on its own merits. The United States has
not even attempted to explain how any alleged errors, omissions or defects
would materially affect the outcome of Mexico's model if "corrected".
Rather, the United States' approach is to avoid engaging with Mexico's proposed
methodology altogether.
Mexico's estimation of export
losses from the impact on prices for Mexican cattle in the U.S. market
9. Mexico's Methodology Paper uses regression analysis to estimate the
economic impact of the COOL measure on the price of Mexican feeder cattle
imported into the United States. The price of Mexican feeder cattle is measured
in New Mexico and Texas and is compared to the price of U.S. feeder cattle
measured at the same locations. Since prices are measured at the same
locations, a limited number of factors explain their difference. Mexico's Methodology Paper
found that the COOL measure depressed exported feeder cattle price by
$0.187/lb.
10. The prices used in the Mexican regression model are measured in the
United States. Once Mexican feeder cattle have crossed the U.S. border and are
at their selling points in the United States, transportation costs and exchange
rates do not matter as these costs are already sunk. Given that these costs
have already been incurred, the price of Mexican feeder cattle is determined
solely by the valuation for feeder cattle by U.S. buyers.
11. Before the COOL measure, U.S. and Mexican born feeder cattle of the
same weight were perfect substitutes. Prices for animals of either origin at
selling points in the United States were essentially the same, as animals of
both origins were physically identical and because animals were not
differentiated by origin in the U.S. supply chain.
12. The COOL measure caused differential treatment of feeder cattle
across origins in the U.S. cattle supply chain. Since the price of Mexican fed
cattle in the United States, which were imported from Mexico as feeder cattle,
is discounted compared to fed cattle of U.S. origin, the prices of Mexican
feeder cattle are discounted. In other words, the discount is passed back from
transactions for fed cattle to transactions for feeder cattle.
13. Mexico's regression model estimates the impact of the COOL measure
as revealed by the price basis. It is well-established in economics that prices
summarize all the information in a competitive market setting. In a supply
chain, this means that a shock at any stage of a supply chain is distributed
through the entire supply chain. Thus all demand and supply shifters that
affect the U.S. cattle and beef industry are summarized in the prices for
feeder cattle. The regression model does not attempt to explain a price-in
level but instead seeks to explain the price basis. Given that both the prices
of U.S. feeder cattle and Mexican feeder cattle summarize the supply and demand
shifters that impact the U.S. beef and cattle supply chain, the price basis
measures what differentially impacts these prices. Because U.S. feeder cattle
and Mexican feeder cattle are identical except for their differential treatment
because of the COOL measure, the price basis regression is able to measure the
total impact of the COOL measure that is distributed down the supply chain onto
the price paid for Mexican feeder cattle. Thus, the relevant variables to
include in the regression model are those that explain the differential value
to U.S. buyers that is distributed down the supply chain rather than all the
variables that explain individual feeder cattle prices. The equations are not "truncated,"
as the United States alleges; rather they include the relevant factors and
exclude those that are not relevant. Accordingly, Mexico's price basis equation
reflects the difference between the price paid for Mexican feeder cattle and
U.S. feeder cattle rather than the "difference between the U.S. price and
the Canadian price" as stated by the United States.
14. The objective of Mexico's regression model is to explain how the
differential treatment of cattle in the United States according to their origin
affected the price paid for Mexico feeder cattle. As explained in the
Methodology Paper and above, Mexico's regression model measures how the COOL
measure differentially impacted the price paid for U.S. and Mexico feeder
cattle as revealed by the price basis. With prices measured in the same
locations, the number of variables that affect the basis is limited. A
regression of the price paid for Mexico feeder cattle in the United States as a
dependent variable with the U.S. price for feeder cattle would require the same
set of explanatory variables as the basis regression presented in Mexico's
Methodology Paper. However, such a regression model would be plagued with
problems that the basis regression does not have. For example, the price of
Mexican feeder cattle exported to the United States contains a unit-root as
shown in Table 1 of Mexico's Methodology Paper. Thus, a conventional linear
regression model would yield biased coefficients unless it is possible to find
a cointegration relationship. The basis does not have a unit-root and thus can
be used as a dependent variable in a linear model.
15. Contrary to the argument of the United States,[7] the Pouliot and Sumner (2014) study does not undermine Mexico's
methodology. Pouliot and Sumner (2014)
are correct in stating that it is theoretically possible that imposing the COOL
measure would result in higher prices in the importing country and lower prices
in the exporting country.[8] The increase in the domestic U.S. price, however, would be small in
practice because the market share of imported cattle is small relative compared
to the total size of the U.S. domestic cattle and beef industry. Moreover, the relevant import volumes change
caused by the COOL measure is an even smaller share of the U.S. market. Further, there is a limited amount of
commingling in the U.S. market, and thus most of the U.S. livestock supply
chain handles single origin animals or meat, keeping costs of compliance with
the COOL measure to a minimum.
16. In fact, a significant increase in prices for products of U.S.
origin in the U.S. livestock supply chain would indicate large compliance costs
and the regulation would not have been adopted if it imposed significant costs
on U.S. firms. Arbitrage between animals
of different origins would cause these prices to differ only by the difference
in cost from the differential treatment associated with the COOL measure. Thus, even though the U.S. domestic price may
marginally increase and the price of the imported animal decrease, the
difference between the two prices reflects exactly the costs associated with
the COOL measure that is passed on to Mexican feeder cattle. Note that even though the United States' EDM
is wrongly specified in many ways, it yields results that are consistent with
Pouliot and Sumner (2014) with the finding that the removal of COOL measure has
a small impact on US domestic prices.
17. The United States' criticisms of Mexico's pricing data are also
unfounded. Mexico's Methodology Paper uses price data collected by the U.S.
Department of Agriculture's Agricultural Marketing Service (AMS). These data
offer an unbiased measure of the price paid for Mexican feeder cattle exported
to the United States and for the U.S. price of feeder cattle in New Mexico and
Texas. The data provided by the AMS are appropriate for this analysis and in
fact the United States used the same data source to calibrate its own EDM.
18. The United States claims that the reports of the U.S. Department of
Agriculture on which Mexico relies for pricing data are from auctions, and
should be considered unrepresentative of Mexican exports. But Mexico did not
use reports from auctions. Rather, it
used the Department of Agriculture's reports on its daily survey of the prices
of Mexican cattle sold in direct sales after crossing the border in New Mexico
and Texas. The quantities of cattle covered by those reports comprise over 70
percent of all U.S. imports of Mexican cattle.
Exhibit MEX-26 contains a statement by the Mexican industry explaining
that Mexican cattle are sold immediately after they have been transferred to
the U.S. side of the border.
19. Also importantly, Mexico's price basis analysis compares prices for
Mexican cattle from Department of Agriculture reports with the prices for U.S.
cattle from Department of Agriculture reports. Mexico's comparisons also
control for the weight and muscle categories available from the Department of
Agriculture reports. This results in the "apples to apples"
comparison that the United States says is so important.
20. Remarkably,
the United States criticizes Mexico's model for relying on actual market data
for the relevant period. In fact, Mexico's Methodology Paper uses a careful,
thorough and state-of-the-art approach. The econometric models include the
relevant sets of control variables to measure the causal impacts of the COOL
measure on prices and quantities.
21. Econometric
modeling is not only a tool for forecasting as the United States asserts, but
is also a well-accepted approach in economics to find causal relationships. A
correctly specified regression controls for the relevant set of variables that
affect a dependent variable. The inclusion of variables that do not pertain to
the economic model and that are correlated with the variables of interest
(dummies for the COOL measure in this specific case) biases the coefficients.
Mexico's Methodology Paper uses a careful approach to include only the
variables that are economically relevant in the regression models.
22. Mexico's regression model includes monthly dummy variables and a
dummy variable for the drought to control for weather effects on the quality of
feeder cattle. For instance, higher temperatures can cause more significant
weight loss over the long distance.
23. Macro-economic variables (e.g. unemployment and GDP) and input cost
variables are not relevant because the regression model compares the price of
two substitute goods for which demands are affected by the same shocks. Input
costs previously incurred are not relevant as they are already sunk costs at
the moment when feeder cattle are sold.
24. Changes in quantities are not relevant in the regression model
because the model compares prices for two substitute goods measured at the same
locations that are impacted by the same demand-side variables. U.S. production
volumes and Mexican export volumes of feeder cattle have nothing to do with how
U.S. feeder cattle buyers differentiate the value of feeder cattle of different
origins.
25. Increase in Mexican beef processing and beef exports are irrelevant
to the difference in the U.S. prices paid for U.S. born feeder cattle and
Mexican born feeder cattle. The decision to export Mexican feeder cattle has
already been made once feeder cattle cross the U.S. border. As the regression
model compares prices for feeder cattle of two origins at the same locations,
it is their relative values to U.S. buyers that determine the difference in
their prices. Mexican beef processing and beef exports have nothing to do with
how U.S. feeder cattle buyers differentially value feeder cattle of different
origins.
26. Mexican feeder cattle exported to the United States meet all animal
disease requirements for exports to the United States. Although in the past
animal diseases have had market impacts, they do not affect U.S. feeder cattle
buyers' differential valuation of feeder cattle of different origins and thus
should not be included in the regression model.
27. The data used in Mexico's Methodology Paper are monthly, and monthly
dummy variables control for the effect of U.S. holidays, if any.
28. Including variables that do not have a causal effect on the relative
price of U.S. and Mexican feeder cattle prices would bias the coefficients for
the COOL variables if these variables are correlated with the COOL variables.
Thus, the regression model used in Mexico's Methodology Paper includes only the
variables that are relevant to explain the difference between the price for
U.S. feeder cattle and the price of Mexican feeder cattle, both measured by
sales within the United States.
29. Contrary to the argument of the United States, all the variables to
explain the difference in the valuation of U.S. buyers of feeder cattle from
the United States and Mexico are included in the regression model. As explained
in the Methodology Paper and above, macro-economic variables and other
variables that can impact the livestock industry have nothing to do with the
difference in value that U.S. buyers assign to feeder cattle of different
origins. The United States asserts that other variables are relevant, but does
not explain why.
Impact of the COOL Measure on Export Volumes
30. To
estimate the impact of the COOL measure on the export of feeder cattle, Mexico's
methodology uses a small simulation model. A simulation is the most appropriate
approach to estimate the loss in export volume because there are causal
variables that are missing to explain the export of feeder cattle. Notably, a
variable to measure expectations about the drought does not exist. The model
found that the COOL measure suppressed export volumes of Mexican feeder cattle
by 342,476 for 2014.
31. The
simulation is based on the estimated impact of the COOL measure on the price
that is estimated econometrically. Given that the export price for Mexican
feeder cattle is suppressed, there is a corresponding decline in the export
volume of feeder cattle. This decline in the export quantity is measure based
on calculated export supply elasticity that is based on elasticity estimates
from the literature and a carefully measured export share of Mexican feeder
cattle.
32. The
United States attacks Mexico's export supply elasticity. But the export supply
elasticity for feeder cattle exported to the United States is derived based on
observed data in a transparent way in Mexico's Methodology Paper. The value of
4.0 for the export elasticity is very reasonable given the size of Mexican
cattle market, the structure of Mexican cattle market and empirical evidence on
supply and demand elasticities provided in Marsh (2003), among others. An
elasticity of 4.0 is a conservative estimate that is consistent with observed
data, the presented empirical evidence and the length of run over which the
market adjusts to the introduction or the removal of COOL measures. The
approach employed by Mexico is transparent and consistent with economic theory,
unlike the elasticity employed by the United States in its EDM which considers
the wrong length of run and employs an elasticity estimated for a completely
different product, wholesale meat.
33. The
United States also criticizes Mexico's suggested level of nullification or
impairment based on its relative size compared to the current value of Mexican
feeder cattle exports to the United States. The large relative value is not
surprising given that the basis used to calculate the relative value, trade
volumes and prices are depressed under the COOL measure. As an example of why
the comparison offered by the United States is incorrect, if the volumes of
trade under COOL had fallen to zero, then the relative size would have been
infinite. Mexico notes that the United States alleges that it is incredible
that Mexico could export 30 percent more feeder cattle. But Mexico's 2014
exports were 1,115,855 head, while its exports previously have been as high as
1,653,408 in 1995 and more recently in 2012 – before the amended COOL measure
was implemented – were 1,468,189.
It is therefore completely realistic that the Mexican industry can increase the
quantity of its exports by 30% over the 2014 figures.
34. The
United States' criticisms of Mexico's calculation of the impact of the COOL
measure on Mexican exports of feeder cattle to the United States are both
superficial and illogical. Equation (5) in Mexico's Methodology Paper (relating
to the quantity of exported cattle) is the same as equation (31) in
Exhibit US-4 that describes the United States' EDM. The single equation is
sufficient and does not need to account for the complexity of the feeder cattle
market in Mexico and the United States because, as explained previously, this
is accounted for in the estimated coefficient of the impact of the COOL measure
on the price of Mexican feeder cattle exported to the United States in the
price basis regression. The United States attempts to mix the different
analyzes together in a confusing and incorrect manner.
35. Exports
of livestock from Mexico and Canada to the United States are significant but
nonetheless represent a small share of the total U.S. livestock market (2013 Final Rule, 78 Fed. Reg. at 31367).
Changes in export volumes from Mexico and Canada would thus have a small impact
on U.S. livestock prices. Furthermore, the United States is a very large
country and exports of cattle from Mexico are made to a very different area
than the exports from Canada, thus limiting direct competition between Mexican
and Canadian cattle.
The United States' EDM is Unreliable
36. The
United States argues that only an EDM is appropriate for use in evaluating the
impact of the COOL measure on Mexico, but that is incorrect. The relevant economic literature – e.g.,
peer-reviewed economic journals that focus on applied economics – confirms that
econometric analysis is the standard approach for ex post
evaluation of policy programs. Importantly, the U.S. government itself has
collected and published most of the relevant data needed to estimate the impact
of the COOL measure as used in Mexico's Methodology Paper. Therefore, there is no lack of reliable data
needed to estimate the impacts of the COOL measure and no reason not to use it.
37. The
United States also argues that econometric analyzes are not favored in WTO
disputes, but overlooks that the Panel in US – COOL found
that an econometric model provided robust evidence that the COOL measure had a
negative and significant impact on Canadian imports shares and price basis.
38. The
United States' EDM suffers from a number of deficiencies. In particular, the
U.S. EDM is calibrated using short-run elasticities, while the full impact of
the COOL measure can only be measured with a long-run economic analysis. The
use of a short-run analysis grossly underestimates the impact of the COOL
measure.
39. Also,
the structure of the U.S. EDM is fundamentally flawed. Mexico's Exhibit MEX-29
provides a schematic of the structure of the U.S. model. The U.S. analysis
assumes that all cattle in the United States are identical, so that the same
costs of the COOL measure are incurred as products move through the supply
chain regardless of the country of origin. The model ignores costs from
segregating livestock and meat according to the country of origin which have
been recognized by the panel in earlier rulings.
40. For
the correct application of the labels at retail, cattle and meat must be
segregated according to their origin. A correct structure for an EDM requires
taking into account that cattle of different origins be kept separated and with
different costs of the COOL measure according to the cattle's country of
origin. An illustration of such a structure is provided on page 4 of Exhibit
MEX‑29.
30. Moreover,
the United States bases its calculation on export prices and volumes that are
suppressed by the COOL measure. A correct calculation of impairment is one
where the baseline is the situation where the COOL measure has not yet been
implemented and then the COOL measure is introduced. Mexico's methodology follows this correct
approach.
41. Instead
of using the available data to directly estimate the impact of the COOL measure
on the export of Mexican feeder cattle to the United States, the United States
proposed a fictional world that is examined within an EDM and proposes to use
this fictional world for the calculation of nullification and impairment. The modeling
assumptions in the EDM presented by the United States assume that there has
been no denial of equal competitive opportunities for Mexican cattle, which
conflicts with the findings of the Panel and the Appellate Body.
42. Specifically,
a complex EDM such as the one proposed by the United States relies on a large
number of assumptions that result in a poor approximation of the causal effects
of policy changes. In particular, the United States' EDM ignores the specifics
of the segregation technology utilized by firms to comply with the COOL measure
and ignores corner solutions, while assuming there is perfect competition in
the US livestock industry. These modeling flaws cause the United States' EDM to
grossly underestimate the impacts of the COOL measure. The United States' EDM
oversimplifies market conditions and is inconsistent with market realities with
respect to the path that the U.S. livestock industry has taken to comply with
the COOL measure.
43. An
EDM cannot be an appropriate substitute for econometric analysis to evaluate
the impact of policy changes ex post. Actual
changes in market prices and quantities reveal the true impact of policy
changes given existing market realities. Unlike a supply chain model such as
the one offered by the United States, econometric analysis does not rely on
assumptions about market structure and model calibration and instead allows the
data itself to reveal impacts on prices and quantities. Data to evaluate the
market impacts of the COOL measure, as demonstrated in Mexico's Methodology
Paper, are readily available and can be used to estimate the causal impacts of
the COOL measures.
44. Even
if one were to accept that an EDM is appropriate to apply in these
circumstances (ex post), there are a number of
specific problems with the United States' EDM.
Tonsor et al. (2015) made a proper use of an EDM in their ex ante analysis of the costs of the COOL measure.[9] This is the type of
analysis that is typically performed using EDMs, unlike the United States ex post analysis. The EDM of Tonsor et al. (2015) does not
specifically consider imports but does account for the segregation costs as
described in the report prepared by Informa Economics on the impact of the COOL
measure by using the weighted average of the costs associated with the COOL
measures for firms that source animals from a single origin and firms that
source animals from multiple origins.
45. The
United States' EDM is inconsistent with the fact that processing plants
accepting imported animals pass the cost of the COOL measure onto the price
they pay for animals that were previously imported as feeder cattle. The Panel in the underlying proceedings made
findings on this issue:
In fact, there is direct
evidence of major slaughterhouses applying a considerable COOL discount of USD
40-60 per head for imported livestock.
This proves that major processors are passing on at least some of the
additional costs of the COOL measure upstream to suppliers of imported
livestock. We have no evidence of a similar discount being applied to suppliers
of domestic livestock, nor has the United States responded to the evidence
submitted by Canada and Mexico in this respect.[10]
46. Also
in the original proceedings, the Panel undertook a detailed examination of the
impact of the COOL measure on imported livestock, and found that competitive
opportunities were reduced in at least sixteen significant ways, namely:
a) a
considerable COOL discount being applied by several major processors to
imported livestock and the absence of a similar discount being applied to
domestic livestock,
b) plants
and companies simply refusing to process any imported livestock,
c) fewer
processing plants accepting imported livestock,
d) certain
suppliers having to transport imported livestock longer distances,
e) plants
processing imported livestock at specific limited times, namely on specific
days of the week or only after specific hours of the day,
f) additional
logistical problems and additional costs for certain imported livestock
suppliers,
g) due
to congestion resulting from limited specific-time deliveries, certain imported
livestock suppliers faced increased difficulty in obtaining delivery trucks or
using trucks in the most efficient way,
h) transportation
delays for certain suppliers of imported livestock,
i) increased
transportation costs for certain suppliers of imported livestock,
j) less
efficient transportation for certain suppliers of imported livestock because of
fewer deliveries due to the longer distance and less turn-around time,
k) changes
to contractual terms for suppliers of imported livestock to incorporate a COOL
opt-out clause to allow processors to unilaterally terminate or amend their
contracts with suppliers of imported livestock,
l) cancellation,
termination or non-renewal of supply contracts for imported livestock,
m) replacement
of long-term contracts with spot contracts at lower purchase prices,
n) 14
days advance notice being required for suppliers of Mexican cattle at various
U.S. processing facilities,
o) certain
suppliers of domestic livestock suffered significant financial disadvantages
due to price discounts for imported livestock as a result of the COOL Measure,
and also due to the refusal of financial institutions to provide credits and
loans to Canadian livestock producers because of the risks resulting from the
COOL Measure, and
p) exclusion
of imported cattle from premium beef programs which are particularly profitable
for livestock suppliers.[11]
47. In
addition, the Informa Study clearly describes that firms that handle single
origin animals/meat incur substantially smaller costs of compliance with COOL
than firms that deal with animal/meat from multiple origins. The United States'
EDM is therefore inconsistent with the actual cost structure and impact of the
COOL measure.
48. Moreover,
equations (18) to (23) of the United States' EDM in Exhibit US-3 are based on
an implausible assumption that removal of the COOL measure will have the same
impact on the prices of imported livestock as on the prices of U.S. livestock
of the same weight categories. This is inconsistent with the pattern of
discrimination found to exist under the COOL measure. Under the COOL measure,
animals of different origins are imperfect substitutes. The COOL measure
requires that feeder cattle imported from Mexico be differentiated from U.S.
born cattle at later stages of the supply chain, so that the meat from these
animals can be correctly labeled according to their origins as specified by the
COOL measure. The requirement to differentiate animals according to their
origins impose additional costs that can be averted by using animals of a
single origin, which is precisely why the COOL measure has a differential
impact in the price of imported Mexican cattle.
Equations (18) to (23) simply assume away this reality.
49. The
United States' EDM is set up in such a manner that the results regarding
livestock prices are given by what the United States calls the "import
wedge." For instance, in tab 16 (Complete Results) of Exhibit US-3, the
change in the imported Mexican feeder cattle price in the United States is USD
12.63 while the change in the Imported Mexican feeder calf price in Mexico is
USD 14.88. The difference between these two values is USD 2.25, which is the
value of what the United States calls the "import wedge" for "farrowing
and cow calf calves from Mexico" in tab 13 (COOL costs) in Exhibit US-3.
This feature of the EDM is inconsistent with the fundamental economic theory
that the costs of a policy program are distributed through the whole supply chain. It is also inconsistent with established
facts that all the cost of the COOL measure for the entire supply chain are
passed onto the price of imported animals, not just the cost of the COOL
measure associated with the imported animals. Moreover, it is inconsistent with
the evidence established during the proceedings that, even before the
implementation of the amended COOL measure, processors were imposing a much
higher "COOL discount" of up to USD 40 to 60 per head. Indeed, the
Panel in the original proceedings observed that "[i]n the absence of a
large share of US consumers willing to pay a price premium for country of
origin labelling, the cheapest way to comply with the COOL measure is to
process only US-origin livestock, all other things being equal", that the "other
possibility is to continue processing imported livestock through segregation,
which entails additional costs in virtually all cases" and that "[e]ither
process configuration is likely to cause a decrease in the volume and price of
imported livestock".[12]
50. Another
problem with the United States' EDM is that the set of elasticities utilized
are inappropriate to measure the full impacts of withdrawing the COOL measure.
Incorrect elasticities values contribute to a severe underestimation of the
impacts of the COOL measure on imported feeder cattle from Mexico. The United
States uses elasticities from previously published works, but those studies had
a very different objective than measuring nullification or impairment and a
different length of run. In some cases, the United States uses elasticities for
a completely different product. The United States' model uses some of the
elasticities reported in Tonsor, et
al. (2015) to derive short-run estimates (one-year). But complete removal of the COOL measure
would require a period of adjustment that exceeds one year. The inappropriate
length of run of the United States' model causes underestimation of the market
impacts of the COOL measure.
51. Moreover,
the United States' EDM uses the same elasticity of supply for U.S. fed and U.S.
feeder cattle. There is no economic rationale for this assumption and this is
not supported by the work of Tonsor
et al. (2015).
52. The
United States set the export supply elasticity of Mexican feeder cattle at 1.83
to equal the supply elasticity for U.S. imports of wholesale meat. However, there is no economic rationale to
set the export supply elasticity of feeder cattle to be the same as for
wholesale meat. The Mexican feeder
cattle export supply elasticity depends on the Mexican domestic demand and
supply elasticities and the export share. There is no economic rationale for
the export supply elasticity of feeder cattle to equal the export supply
elasticity of meat. The supply and demand conditions for these two products are
significantly different.
53. In
its EDM in Exhibit US-3, the United States uses cost estimates from the
Regulatory Impact Analysis (RIA) prepared by the Agricultural Marketing Service
of the U.S. Department of Agriculture, which were prepared in connection with
the 2009 and 2013 versions of the COOL regulations (Exhibit US-1 and Exhibit
US-2). The RIA is a costs and benefits
analysis that is concerned with regulatory impacts on the United States'
economy. The 2009 and the 2013 RIA
therefore mainly focus on labelling and completely ignore segregation costs
that are the source of the differential treatment of imported livestock. The
United States accordingly ignores in its model a significant source of costs,
which leads to severe underestimation of the amount of nullification and
impairment. As an alternative, the
United States could have used
the Informa Study as an objective and unbiased source of information about the
costs of the COOL measure. The estimated
costs in the Informa Study are substantially higher than those in the RIA, in particular for firms that source
livestock from more than one country of origin. The EDM presented
by the United States fails to account for the cost of segregation and
segmentation described in the Informa Study.
54. The
United States' EDM also includes an adjustment for the United States' assertion
that costs should be reduced to account for the exemptions from the COOL
measure, such as for processed products. But the Panel in the underlying
proceedings found that the burdens of the COOL measure were imposed on all
imports of Mexican cattle, because at the time of importation the purchasers do
not know their ultimate use. The United States pursued this argument with the
Appellate Body and it was rejected. Accordingly, there is no justifiable basis
for including such an adjustment in the United States' model.
Impact on Mexican Domestic Market Prices
55. Mexico's
Methodology Paper established that the COOL measure's impact on Mexico's
exports of cattle has had effects not only on Mexican exports, but also on
sales of cattle in Mexico's domestic market. The impact of the domestic price
suppression is calculated to be USD 198 million. Like the export price and
volume effects, this domestic price suppression effect is directly related to
the nullification or impairment at issue. In simple terms, the relevant benefit
accruing to Mexico is the right of not having to face a measure like the COOL
measure. By virtue of the nullification or impairment of this benefit by the
COOL measure, Mexican domestic prices have been suppressed. It would
fundamentally undermine the balance of concessions in the WTO Agreements if the
full extent of the benefits accruing to WTO Members were not recognized in this
arbitration. The COOL measure has disrupted the previously integrated North
American cattle market. The price suppression in the Mexican market is a direct
result of this disruption. In legal terms, the price suppression is the direct
effect of the nullification or impairment of the benefits accruing to Mexico
under the WTO Agreements.
56. The
United States argues that, as a matter of law, the harmful economic effects of
price suppression in the Mexican domestic market for cattle cannot be
considered at all. It claims that prior arbitrators have found that they must
only consider trade flows, and seeks to characterize Mexico's quantification of
the harm from domestic price suppression as "some broader, subjective
measure of the overall economic impacts supposedly related to non-compliance."[13] But Mexico has not claimed that it should be
compensated for general effects in its domestic economy; rather, it has shown
the causal effects of the COOL measure specifically on the Mexican cattle
industry. Moreover, the covered agreements contemplate that nullification and
impairment can have indirect effects, and prior arbitrators have not excluded
the possibility that effects in domestic markets can be taken into account.
These points are explained in detail below.
57. In evaluating claims for
economic harm, arbitrators in prior disputes have focused on whether there was
a sufficient "causal link" between the measure at issue and the
alleged harm, such as in in EC – Hormones (US)
(Article 22.6 – EC). The
requirement of a causal link between benefits being nullified and impaired and
the measure at issue derives from the language of GATT Article XXIII:1, which
establishes that a nullification or impairment must arise "as the result
of … the failure of another contracting party to carry out its obligations".
58. The concept of "causal link"
has been discussed in cases arising under other agreements. For example, in US –Steel Safeguards, the Appellate Body, in the context of
examining the term "as a result of" in Article XIX:1(a), found that
there is a need to establish a causal link between increased imports and
unforeseen developments when imposing a safeguard measure:
Turning to the term "as
a result of" that is also found in Article XIX:1(a), we note that the
ordinary meaning of "result" is, as defined in the dictionary, "an
effect, issue, or outcome from
some action, process or design". The increased imports to which this
provision refers must therefore be an "effect, or outcome" of the "unforeseen
developments". Put differently, the "unforeseen developments"
must "result" in increased imports of the product ("such product")
that is subject to a safeguard measure.[14]
59. Similarly,
in US – Wheat Gluten, the Appellate Body
stated:
The word "causal"
means "relating to a cause or causes" while the word "cause",
in turn, denotes a relationship between, at least, two elements, whereby the
first element has, in some way, "brought about", "produced"
or induced" the existence of the second element. The word "link" indicates simply
that increased imports have played a part in, or contributed to, bringing about
serious injury so that there is a causal "connection" or "nexus"
between these two elements.[15]
60. This
case does not involve a situation in which economic harm is based on
speculation and/or is not capable of quantification. Mexico has demonstrated an
extremely close causal link between the COOL measure and price suppression in
the Mexican domestic market. In particular, it is undisputed that the Mexican
and U.S. markets for cattle are tightly integrated. The Mexican cow-calf
industry was structured over one hundred years ago to supply the U.S. market.
There are no export markets for Mexican cattle other than the United States.
Mexico's methodology takes into account the loss of exports caused by the COOL
measure, to avoid double-counting.
61. The
only criticism the United States makes of the methodology is to state, vaguely,
that "Mexico does not account for other factors impacting its domestic
sale of livestock that are completely unrelated to the impact of the amended
COOL measure on export volumes. For
instance, Mexico does not account for the drought's impact on the quality or
life span of Mexican cattle."[16] To the contrary, Mexico's
methodology isolates the impact of the COOL measure, and measures the impact in
terms of lowered prices in the Mexican market.
62. The
calculation of the price suppression in the Mexican domestic cattle market
builds on the estimate of the impact of the COOL measure on the price of
Mexican feeder cattle exported to the United States. The U.S. and Mexican
cattle industries are highly integrated and it is a natural outcome of economic
forces that a shock on the price of Mexican feeder cattle exported to the
United States to be transferred onto prices in the domestic Mexican cattle
market. The approach in Mexico's Methodology Paper is to estimate a price
transmission regression to measure price linkage between the U.S. feeder cattle
market and the Mexican feeder cattle market. It is found that in the long run, $0.678/lb of a $1/lb increase in the price of feeder
cattle exported to the United States is transmitted to the Mexican domestic
price of feeder cattle. Thus, given that the COOL measure depressed exported
feeder cattle price by $0.187/lb, the corresponding price suppression to the
Mexican domestic price is -$0.127/lb. Applying the price suppression to the
Mexican domestic feeder cattle market yields a total prices suppression loss to
Mexico of $198,628,204.
63. The
calculations of the price suppression loss to Mexico focus on the feeder cattle
market. The losses however certainly extend to the whole Mexican domestic
cattle market. On a per pound basis, the losses to other cattle categories are
smaller and more difficult to measure. However, applied to the whole Mexican cattle
industry, these losses are certainly substantial. Hence, the calculations
offered by Mexico, which focus on the feeder cattle market, should be
understood as a lower bound of the total damage caused to the Mexican cattle
industry by the COOL measure.
64. In
Mexico's view, the price suppression in the Mexican market is a direct effect
of the COOL measure. However, to the extent it might not be direct, it would
fundamentally undermine the balance of concessions in the WTO Agreements if the
full extent of the benefits accruing to WTO Members were not recognized in this
arbitration. Thus, when examining the nullification or impairment of benefits
under the provisions of all of the WTO Agreements including under Article 22 of
the DSU, it is essential to interpret those benefits broadly so that they
include obvious direct benefits but also benefits that are less direct but are
nonetheless real. In this instance, the price suppression in the Mexican market
is undoubtedly real and the benefits accruing to Mexico under the WTO
Agreements should have prevented this price suppression from occurring but have
not because they have been nullified or impaired.
65. For
both violations of the TBT Agreement and the GATT 1994, a Member may bring a
dispute if it considers "any benefit" accruing to it "directly
or indirectly" is being nullified or impaired "as the result of"
a violation by another Member of its WTO obligations. The "nullification or impairment"
referred to in DSU Article 22.4 is the nullification or impairment of "benefits"
accruing to a member in the sense of GATT Article XXIII:1. In other words, the
benefits may be direct or indirect benefits.
66. The
United States nonetheless argues that the Arbitrator may only consider "trade
flows," claiming that arbitrators in past cases have found that the level
of nullification and impairment must be based exclusively on the impact on
trade flows. It cites to the EC – Hormones, US – Gambling, and EC – Bananas
cases in support of this claim. However, the United States is wrong to suggest
that the arbitrators in prior disputes have found that the effects of measures
on domestic markets are excluded from consideration. For example, the arbitrator in US – Byrd Amendment stated:
We do not agree with the
United States that nullification or impairment is to be limited in all
instances to the direct trade loss resulting from the violation. We agree with
the Requesting Parties that the term "trade effect" is found neither
in Article XXIII of GATT 1994, nor in Article 22 of the DSU. Previous
arbitrators' decisions based on direct trade impact are not binding precedents.[17]
67. The
arbitrator in that case further commented that "[t]he use of direct trade
effect in most cases reflects the fact that trade loss is generally more
directly identifiable and quantifiable and that, in such a context, arbitrators
preferred to rely on verifiable figures",[18] and also stated that "…
the ‘trade effect' approach has been regularly applied in other Article 22.6
arbitrations and seems to be generally accepted by Members as a correct application of Article 22 of the DSU"
(emphasis original).[19] In other words, it is
not the only correct application of Article 22.
68. The United States also cites to EC – Bananas (US) (Article
22.6 – EC). In that case, the United
States included a claim for economic harm arising from lost exports to third
countries of inputs, such as fertilizer, for the third countries' production
of bananas. The arbitrator rejected this claim, stating "[t]o the extent
the US assessment of nullification or impairment includes lost US
exports defined as US content incorporated in
Latin American bananas (e.g. US fertilizer, pesticides and machinery
shipped to Latin America and US capital or management services used in banana
cultivation), we do not consider such lost US exports
for calculating nullification or impairment in the present arbitration
proceeding between the European Communities and the United States".[20] (emphasis original)
69. In
this regard, the key point for the Panel was its view that the third countries
could make their own claims for nullification or impairment based on effects on
their own exports of bananas to the EU, and that it was inappropriate for the
United States to make, in effect, a claim of harm based on the harm to those
other countries' exports. Mexico is not claiming harm from loss of exports to
third countries. Furthermore, third countries could not make claims for
nullification or impairment based on the price suppression effects caused by
the COOL measure on Mexican cattle in the Mexican domestic market. Thus, the
arbitrator's findings in EC – Bananas do
not support the United States' argument that indirect effects can never be
considered.
70. In US – Gambling, Antigua argued that losses to the Antiguan
remote gambling industry led to additional losses in other sectors of the
economy, including lower income and government revenues. It characterized these
as indirect effects that should be taken into account in addition to direct
trade effects. To compensate for those effects, it proposed to apply a "multiplier"
to its calculation of the trade effects to arrive at an approximation of the
indirect effects. The arbitrator rejected the multiplier, finding that "the
use of a multiplier reflecting the aggregate change in output for a unit change
in demand would be contrary to some of Antigua's other arguments concerning the
limited impact of remote gambling revenues on GDP."[21] Unlike
Antigua in the US – Gambling case, Mexico has
not made a claim of harm from negative effects to its general economy but
rather has identified concrete effects on the Mexican cattle industry arising
from the COOL measure. Mexico has not made other arguments that are contrary to
its position.
71. In
the EC – Hormones (Canada) (Article 22.6 – EC)
determination cited by the United States, Canada did not propose to include any
economic harm except that to trade flows, so it is unclear why the United
States believes that determination is a relevant precedent. In EC – Hormones (United
States) (Article 22.6 – EC), the arbitrator rejected a U.S. effort
to include an estimated amount for additional exports of edible beef offal that
would have been made as a result of U.S. marketing and promotional efforts that
would have taken place but for the hormone ban. In other words, there had been
no such marketing efforts. The arbitrator declined to take those projected
exports into account on the basis that they were too speculative. The effects on the Mexican domestic market for cattle, in contrast,
are not speculative; they are quantifiable and have a close causal connection
to the COOL measure.
72. The
United States cites to a recent proposal to amend the DSU as evidence that the
DSU currently prohibits the inclusion of domestic price suppression. The
proposal, contained in TN/DS/26, is to amend the DSU to expressly allow the
level of nullification and impairment to include an estimate of the effect of
the inconsistent measure on a country's economy as a whole. Mexico has
not proposed to include in the level of nullification and impairment the impact
of the COOL measure on Mexico's economy as a whole; rather, Mexico has included
the effects of the inconsistent measure on the Mexican market for cattle –
effects that have a close causal link to the U.S. measure. Accordingly,
TN/DS/26 is not relevant to this dispute.
73. The
United States also puts forth an argument that there is a requirement to
evaluate the effect of Mexico's proposed suspension of benefits on the U.S.
economy as a whole, claiming that "[t]he corresponding level of suspension
would need to be decreased by an appropriate calculation of the broader
economic effects on the U.S. economy of the suspended trade."[22] The United States cites
no provision of the covered agreements or any prior arbitration ruling in
support of this argument, because there is none. The level of nullification and
impairment is not measured in terms of its impact on the country that maintains
the inconsistent measure. The appropriate measure is the value of the denial of
benefits, direct and indirect, to the complaining member.
ANNEX
B-3
Executive
summary of the arguments of the United States
I. Introduction
1. Both Canada and Mexico calculate the level of
nullification or impairment as the sum of "export revenue losses" and
domestic "price suppression losses."
In the first instance, these estimates dwarf the historical and current
export value of livestock and in no way reflect the "benefit"
impaired by the amended COOL measure.
Specifically, Canada and Mexico are arguing that if the amended COOL
measure were withdrawn, their exports of livestock would increase 92 percent
and 70 percent, by value respectively, to never before seen levels, and even as
overall demand for beef and pork muscle cuts in the United States has been in
decline since 2008 – with no sign of rebound.
In the second, the claimed "price suppression losses" are not
part of the level of nullification or impairment of benefits accruing under the
Agreement on Technical Barriers to Trade
("TBT Agreement") or the General Agreement on
Tariffs and Trade 1994 ("GATT 1994").
2. In response, the United States explains why the
econometric calculations of the requesting parties produce highly inflated
levels of nullification and impairment.
In contrast to the flawed methodologies proffered by Canada and Mexico,
the United States puts forward a type of partial equilibrium model, which more
accurately estimates the trade effects of the COOL measure, as amended, in the
context of the complex North American market.
Specifically, an equilibrium displacement model ("EDM") is the
most suitable tool for estimating the trade effects of the amended COOL
measure. And
finally, the United States has explained why the requesting parties' claims for
non-trade related damages cannot succeed.
II. Appropriate Calculation of the
Level of Nullification or Impairment
3. Pursuant
to Article 22.6 of the Understanding on Rules and
Procedures Governing the Settlement of Disputes ("DSU"),
the United States objected to Canada and Mexico's proposed levels of suspension
of concessions or other obligations because each party has submitted a proposed
level of suspension that is far in excess of the level of nullification or
impairment attributable to the measure at issue. Article 22.4 of the DSU is explicit and
requires that the "level of suspension of concessions or other obligations
authorized by the DSB shall be equivalent to the level of nullification or
impairment." The requesting parties'
calculations suffer from conceptual flaws and methodological errors that result
in grossly inflated estimates of the levels of nullification or
impairment.
4. In this proceeding, Canada and Mexico have each
gone far beyond an "equivalent" level of nullification in offering a
two-part asserted level of nullification or impairment, which in the first
instance exceeds all possible trade effects, and which in the second instance
is not attributable to the nullified or impaired benefit. As to the former, Canada and Mexico attempt
to quantify the "export revenue losses" attributable to the amended
COOL measure, i.e., the volume and value of
livestock that would have been exported "but for" the amended COOL
measure. The methodologies employed to
estimate the quantity and value effects of the amended COOL measure are
fundamentally flawed and result in requests for levels of suspension of
concessions that are unsupportable. As
to the latter, Canada and Mexico argue that domestic "price suppression
losses" should also be included in the total level of nullification or
impairment. Even if this "loss"
level was determined through a clear and rational methodology, which it is not,
the alleged effects on domestic price are not trade effects and do not relate
to the "benefits" accruing under the relevant covered agreements (the
TBT Agreement and the GATT 1994) that are being nullified or impaired.
5. The DSU does not prescribe any particular way
to demonstrate that the level of suspension requested by each party is
excessive in light of the requirements of the DSU. The United States has established its prima facie case in three different, and independent,
ways. First, the United States has
provided a methodology – the EDM – that more accurately estimates the level of
nullification and impairment than the one proposed by the requesting
parties. Second, the United States has
explained why, even aside from the EDM approach, the somewhat differing
econometric calculations of the requesting parties produce highly inflated
levels of nullification and impairment.
Third, the United States has explained why the requesting parties'
claims for non-trade related damages – i.e., their
claims regarding domestic "price suppression losses" – are legally
invalid.
A. Applied Economic Analysis Is
Necessary to Accurately State the Level of Nullification or Impairment in the
North American Livestock Industry
6. To calculate the amount of nullification or
impairment, one must compare on a prospective basis the imports of the relevant
livestock from Canada and Mexico under the amended COOL measure to the imports
that would occur were the amended COOL measure withdrawn. And to make that comparison, one would look
at the actual relevant U.S. livestock imports during the most recent period
(actual situation), and then estimate the relevant imports of livestock that
would exist during the same period if the amended COOL measure were withdrawn
and all other factors were held constant (the counterfactual).
7. Recognizing these challenges, and the complexity
of the North American livestock markets, the United States uses a type of
partial equilibrium model, an EDM, to estimate the prospective trade effects of
coming into compliance with the DSB recommendations and rulings through withdrawal of the amended
COOL measure. This model compares a
baseline of 2014 trade data to what would happen to supply and demand across
all three countries if the amended COOL measure were withdrawn.
1. Overview of the Equilibrium Displacement Model
8. EDMs are a well-accepted and widely used type
of partial equilibrium model used for applied economic analysis, particularly
in the agricultural sector. In
particular, EDMs are well accepted by economists, and have been widely used in
the economic literature to model and measure the impact of policy changes in
the agricultural sector. In the context
of COOL, the United States notes that there have been at least three
significant studies of the U.S. livestock market using EDMs.
9. Further, prior arbitrators in Article 22.6
proceedings have in the past relied on partial equilibrium or stimulation
models similar to the EDM proposed by the United States. In this regard, the United States notes that
the arbitrator in US – CDSOA (Article 22.6 – US)
considered that where "evaluating the trade effects of the scheme cannot
be accomplished with mathematical precision," "economic science
allows for the consideration of a range of possible trade effects with a
certain degree of confidence." That
is, the use of well-supported and reasoned economic models that recognize the
varying effects of a measure, as the EDM does, has been an important tool for
arbitrators.
2. Explanation of the Equilibrium Displacement Model for the
U.S. Cattle/Beef and Hog/Pork Sectors
10. The United States uses an EDM in order to
estimate the difference between the value of trade flows in 2014 and a
counterfactual situation where compliance with the DSB recommendations and
rulings is achieved. The EDM is a series
of linearized equations that provide economic estimates of the trade shifts
that would occur if the amended COOL measure were withdrawn.
11. The EDM utilizes a multi-animal (covering
cattle/beef and hogs/pork), and multi-sector (representing five levels of the
beef and pork marketing chain), structure.
For each species and at each level, the model establishes baseline
quantities and prices, and then estimates the price and quantity changes due to
an external "shock."
12. In this case, the shock is the immediate elimination
of the amended COOL measure and its associated compliance costs, which appear
in the first four marketing levels. All
other independent variables are held constant at their 2014 levels. In this context, the resulting quantities and
prices are endogenous variables, meaning they are determined within the EDM by
a set of exogenous and computed components.
Exogenous components include the baseline quantity and prices, demand
and supply elasticities, and COOL compliance costs.
3. 2014 Baseline Quantities and
Prices
13. The EDM's baseline utilizes 2014 market
quantities and prices sourced from the U.S. Census Bureau trade data. The most recent full year data reflects all
current market conditions such as transport costs, feed costs, exchange rates,
ownership structures, Canadian and Mexican domestic policies, and environmental
factors as they existed in 2014. The
year 2014 thus provides the most appropriate baseline for the purposes of
determining the nullification or impairment of benefits accruing to Mexico and
Canada under the TBT Agreement and the GATT 1994 on a prospective basis.
14. Construction of the 2014 baseline, as well as
the EDM, depends on certain additional assumptions. The EDM assumes that all marketing levels are
in perfect competition. The EDM utilizes
"fixed proportions" between inputs and outputs through the marketing
channel. The EDM also assumes that
technologies used in the "value-added" sectors provide a constant
return to scale. The EDM further uses
certain "conversion factors" to translate animal standard-sized
livestock from the number of head of livestock into the retail weight in
pounds. Finally, the conversion factors
and the EDM, more generally, are based on an assumption that fed cattle are
1,400 lbs. and fed hogs are 300 lbs.
4. Multi-Animal, Multi-Marketing
Sector Model Structure
15. To accurately estimate the trade effects of the
amended COOL measure at each level of the marketing chain from farm to
consumer, the EDM explicitly models the five distinct levels of the livestock
market: (1) cow-calf and farrowing, (2)
finishing, (3) packing/wholesale, (4) retail, and (5) consumers. To model the complete and integrated
livestock-to-retail meat market, this model also incorporates imported
livestock from Mexico and Canada, as well as imports and exports of pork and
beef. The model therefore captures the
elements of supply and demand relevant to the livestock/meat market in North
America.
16. The EDM uses four sets of equations, "identity," "price,"
"value-added," and "structural," to define the market and
analyze shifts resulting from withdrawal of the amended COOL measure. These equations are based on the assumption
that equilibrium conditions exist at each stage of production.
5. Explanation of Elasticities and COOL Compliance Costs
17. There are two primary input parameter values
utilized by the EDM: elasticities and
COOL compliance costs.
a. Elasticities
18. The EDM's structural supply and demand
equations are linearized and use the elasticities, consistent with previous
COOL EDM studies, to determine the responsiveness of prices and quantities in
the model to exogenous shocks. As discussed in academic literature and
noted in Mexico's Methodology Paper, data and time constraints render
impractical estimating all supply and demand elasticities econometrically. Therefore, the EDM follows the same approach
as other EDM studies and uses supply and demand elasticity estimates
established in and vetted by peer-reviewed academic literature.
19. The EDM utilizes short-run supply elasticities
for the supply of U.S. feeder animals and the supply of imports of feeder
animals, slaughter animals, and wholesale meat drawn from academic
sources. In this context, short-run is
typically defined as one to two years, while long run is typically defined as
ten years. The EDM also utilizes demand
elasticities for U.S. retail meat (own-price and cross-price elasticities) and
U.S. wholesale meat exports.
20. Previous academic studies of the North American
livestock market do not provide supply elasticities for U.S. imports of feeder
or slaughter animals. The United States
has thus set these elasticities to equal the supply elasticity for U.S. imports
of wholesale meat imports. This is
consistent with the expectation that the import supply elasticities for these
animals would be higher than those for domestic supplies, and is supported by
other studies that developed lower estimates for these parameters. Canada claims, however, that these elasticities
are inappropriate because the ratio of export supply to total supply is
important, and the (alleged) long-run must be calculated on an annual basis (and purports to do so for 2014). Canada provides no clear methodology or data
to support its extreme export supply elasticities (which range from 12.6 to
126.3), which are much higher than those developed by academics specifically
considering the underlying markets.
b. COOL Compliance Costs
i. RIA Cost Estimates
21. To estimate the trade effects of withdrawing
the amended COOL measure, the costs of COOL compliance are estimated and
removed from the EDM at each level of the beef and pork production chain. The COOL cost estimates in the EDM are based
on the Regulatory Impact Analyses ("RIAs") conducted by the U.S.
Department of Agriculture ("USDA") with respect to the 2009 and 2013
COOL final rules. The United States has
also put forward an alternative based on the Informa Economics report costs
which form the far upward bound of likely costs.
22. Although the RIA costs assume that exclusively
U.S.-origin meat and mixed origin meat are subject to the same incremental
direct costs at the farm, finishing, packer and retail levels, differential
impacts arise due to differing elasticities for import supply and domestic
supply. The EDM captures and measures
these differences by imposing appropriate import and domestic supply
elasticities. That is, imported products
are more sensitive to incremental cost increases and reflect these changes more
severely in price and quantity changes.
This difference reflects the differential compliance costs imposed on
Canadian and Mexican livestock suppliers.
23. But for the compliance costs related to the
2009 and 2013 COOL measures, the value of Canadian and Mexican livestock
exports to the United States would have exceeded the 2014 baseline level of
exports. Specifically, Canadian feeder
pig exports would be US$3.75 million higher than 2014 levels, and Canadian
slaughter hogs would have been US$0.35 million higher. Canadian feeder calf exports would have been
US$21.45 million higher and slaughter cattle would have been US$17.64 million
higher. Mexican feeder calf exports
would have been US$49.18 million higher than 2014 export levels.
ii. Informa Economics Cost Estimates
24. Recognizing that the original panel and
compliance panels have found that some portion of U.S. costs may be shifted up
the supply chain and imposed on importers, the United States has also put
forward an alternative based on the Informa Economics report costs which form
the far upward bound of likely costs. As
the original panel noted, however, the "Informa Report is silent on its
methodology and the sample considered (i.e., time
period, geographical zone, number of firms surveyed)," and thus is not "reliable
and precise as regards its exact quantification of the costs of the COOL
measure." These costs in fact
represent an exaggeration of the compliance costs for mixed origin product, and
the far upward bound of potential segregation and compliance costs.
25. Using this cost wedge and assuming that U.S.
retailers and packers will push costs associated with mixed origin animals up
the supply chain, the value of Canadian and Mexican livestock exports to the
United States would have exceeded the 2014 baseline level of exports. Specifically, Canadian feeder pig exports
would be US$62.30 million higher than 2014 levels, and Canadian slaughter hogs
would have been US$5.10 million higher.
Canadian feeder calf exports would have been US$34.30 million higher and
slaughter cattle would have been US$27.01 million higher. Mexican feeder calf exports would have been
US$78.95 million higher than 2014 export levels.
6. Conclusion
26. As demonstrated by the EDM, the more appropriate level of nullification
or impairment is approximately US$43.22 million per year for Canada, and
certainly no more than US$128.71 million per year. With respect to Mexico, the more appropriate
level of nullification or impairment is approximately US$47.55 million per
year, and certainly no more than US$78.95 million per year. This analysis demonstrates that the levels of
suspension of concessions requested by Canada and Mexico are in excess of the
appropriate levels of nullification or impairment.
III. The Levels of Suspension of Concessions or Other Obligations
Proposed by Canada and Mexico Far Exceed the Levels of Nullification or
Impairment
27. The
requesting parties utilize econometric methods that are fundamentally incapable
of estimating the impact of the amended COOL measure in the complex North
American livestock and meat market.
Their "export revenue loss" calculations depend on unrealistic
assumptions and suffer from serious methodological deficiencies that render
their estimates incorrect. As noted consistently
by previous arbitrators, the proposed
level of nullification or impairment must reflect the "benefit"
accruing under the relevant covered agreement allegedly nullified or impaired "as
a result of" the breach found by the DSB. That
is, the proposed level must be an accurate reflection of the trade that would
have occurred "but for" the inconsistent amended COOL measure, and
not a reflection of unrelated market drivers or circumstances.
A. Canada
and Mexico's Proposed "Export Revenue Losses" Methodologies Are
Fundamentally Flawed and Result in Overstatements of the Levels of
Nullification or Impairment
28. The
United States, Canada, and Mexico agree that the "trade effects" of
an inconsistent measure are determined by evaluating the difference between a
baseline annual export value and the estimation of what that export value would
be if the amended COOL measure costs were eliminated. However, neither Canada nor Mexico's alleged
level of nullification or impairment reflects the established patterns of
supply and demand in North America or the realities of the livestock
industry. Canada's total hog and cattle
export value for 2014 was US$1.744 billion.
Canada's estimated level of nullification or impairment, US$1.61
billion, suggests that export revenues would increase by 92.3 percent by value
if the COOL measure was eliminated.
Mexico's total feeder cattle export value for 2014 was US$737
million. Mexico's suggested level of
nullification or impairment suggests that marginal revenue will increase by as
much as 70 percent by value.
1. Econometric Modeling Is Not Well Suited to
Accurately Determining Trade Effects
29. Canada's
Methodology Paper attempts to use linear regression analysis to econometrically
estimate the "reduction in the average weekly exports from Canada to the
United States caused by the amended COOL measure," and the "price
basis." Mexico's Methodology Paper
seeks to determine "price basis" through econometric analysis, but
abandons this methodology when determining the impact of the amended COOL
measure with respect to exports.
30. Econometric
modeling analysis seeks to estimate the statistical relationship between a
variable of interest (the dependent variable) and other explanatory variables
(the independent variables) as a tool for forecasting how changes to those
independent variables would impact the dependent variable. Econometric modeling, however, is not an
appropriate approach for determining the level of nullification or impairment. For example, it is widely understood that
econometric models are dependent on the inclusion and accurate estimation of
exogenous variables, are limited by the ability to incorporate accurate real
world data, and must ensure that the relationship between the variables and
data is accurately identified. Failure
to address these issues will lead the model to attribute to the amended COOL
measure trade effects that are due to some other factor. The concept of "non-attribution" is
one that is familiar under the covered agreements and was addressed by the
recent China – GOES compliance panel. These concerns make econometric models poorly
suited for analyzing complex markets, such as integrated and vertically linked
animal and meat markets, which are subject to numerous and overlapping variables
that may impact the dependent variables.
2. Canada and Mexico's Models Are Mis-specified
Because the Models Omit Numerous Necessary Explanatory Variables
31. The reduced form econometric
modeling proposed by Canada and Mexico is far too simplistic to accurately
isolate and quantify the magnitude of any potential effects of the amended COOL
measure. In particular, Canada and
Mexico's limited analysis does not consider a number of important explanatory
variables impacting the North American livestock and meat markets between 2005
and 2015. Failure to accurately control
for relevant factors results in attributing to the amended COOL measure effects
that are instead due to other factors.
For this reason, Canada and Mexico's proposed levels of nullification or
impairment far exceed the "benefit" being impaired.
32. To
accurately isolate and assess the quantity and price impact of the amended COOL
measure, the requesting parties' models should not choose to include or exclude
explanatory variables based on the bias requesting parties assume the variable
will create or on the assumption that the effect is small – as they have done
in these arbitrations. Rather, all
explanatory variables should be included in the analysis.
33. Specifically,
the requesting parties must effectively control for numerous independent
variables, which also had an impact on quantity and price during this
period. These independent variables
include, but are not limited to:
·
Economic
Fluctuations and Recession: Significant economic fluctuations affecting
the price and quantity of livestock exports to the United States have occurred
during the period used by Canada and Mexico.
The global economic crisis resulted in a worldwide slowing of trade and
an overall contraction of agricultural markets between 2007 and 2009. The recessions had different origins and
impacted each of the three economies differently. The U.S. recession, which lasted between
December 2007 and June 2009, was largely driven by domestic factors in the
housing and banking sectors. Canada
entered economic recession in December 2008, which is a full year after the
United States. Mexico's recession lasted
from October 2008 to March 2009.
Despite addressing the most significant economic downturn in recent
memory in other submissions and academic papers, Canada and Mexico provide no
assessment of the recession's effect on export quantities or the price
basis. Instead, Canada and Mexico
attribute the total effect of the economic downturn to the amended COOL
measure.
·
Increased Feed
Costs: Feed
costs, as one of the single largest input into livestock production, play a
significant role in determining price and trade flows. For instance, when the cost of feed is high,
the profitability of feeding cattle declines, encouraging increased slaughter
or export of animals. Between 2005 and
the present, feed costs in North America have shifted for a number of reasons,
including drought, biofuels policy, changing export demands, and shifting
domestic demand. In fact, feed costs not
only change throughout the period of the amended COOL measure, impacting the
price and quantity of livestock shipped, but feed costs affect Canada, Mexico,
and the United States differently and must be accounted for in econometric
price and quantity equations to ensure that changes in feed costs over time are
not incorrectly attributed to the estimated effects of the amended COOL
measure.
·
Shifting
Transportation Costs: Transportation costs can significantly impact
cattle trade between Canada and the United States, and Mexico and the United
States. When transportation costs, which
are linked to the price of fuel, are high the incentive to ship Canadian cattle
to the United States diminishes.
Therefore, U.S. packers will purchase fewer Canadian livestock and
Mexican cattle, and the price of imported livestock will decline. This is particularly clear as Canada's own
submission specifies differences in transportation costs between costs for
Canadian and U.S. producers. Unless these
costs are properly accounted for, there is no way through an econometric
analysis to precisely isolate the effects of the amended COOL measure on the
price basis.
·
Lingering
Effects of BSE and Other Animal Diseases: The discovery of bovine spongiform encephalopathy
("BSE") in Canada in 2003 has also had lingering effects on the
Canadian market. While Canada has
attempted to account for the trade disruption between Canada and the United
States, it has not addressed the impact of bans enacted by other trading
partners on imports of live cattle, beef, and beef products. Conversely, Mexico continues to benefit from its increased market share
in a number of Canada's prime export markets, which are periodically closed to
Canadian exports due to BSE cases (reported as recently as February 2015).
·
Shifting
Livestock Processing: Both Canada and Mexico have functioning
slaughter and processing sectors which provide meat for domestic consumption as
well as export. The relative health of
this sector and, in particular, shifts in production capacity have a
significant impact on the domestic price of livestock and the competitive
opportunities for Canadian/Mexican farmers and feedlot owners. This should be considered in any econometric
analysis.
·
Weather
Patterns:
Weather related disruptions, such as drought, can significantly impact
export levels. For instance, between
2011 and 2014 a significant drought affected Mexico and the U.S. Southwest. Drought both encouraged exports from Mexico,
and increased slaughter (and a decline in stocks) in the United States. Canada has not controlled for the impact of
this drought or other weather conditions.
Rather, Canada suggests that if this were included in the econometric
model specification the COOL impact would be larger because the drought had
increased demand for imports of Canadian cattle to be used for breeding stock
rather than for slaughter. However,
Canada misunderstands the impact of the drought in the context of the integrated
market. As Mexico indicated, the drought
and expectations regarding its length and cost encouraged Mexican farms to
export to the United States more feeder animals at lower weights and lower
prices. This increased supply from
Mexico decreased demand for Canadian feeder animals, and this effect should not
be attributed to the amended COOL measure.
·
U.S. Holidays: Significant holidays are
often preceded by an increase in demand for beef and pork. But in their Methodology Papers, Canada and
Mexico fail to address the influence of these holidays on quantity impacts or
price basis.
·
Competing
Imports:
Canada does not appear to consider the impact of U.S. or Mexican
production on the ability of Canada to export to the United States, and Mexico
does not consider the impact of Canadian and U.S. production on Mexican
exports. Canada suggests that the United
States is so large that the presence of an additional significant supplier of
feeder cattle is irrelevant. This is
erroneous. Canada further suggests that
imports on the southern border do not affect the prices or quantities imported
on the northern border. This stands in
contrast to Canada's statements regarding the single integrated market, and is
also in error. Failure to include
another significant market player will result in Canada attributing to the
amended COOL measure the impact of factors related to the supply of Mexican
feeder cattle and in Mexico attributing to the amended COOL measure impacts
related to the supply of Canadian livestock.
34. Finally,
a wide variety of factors influence the quantity of livestock crossing the
border and the price at which the livestock is sold, and because Canada and
Mexico are seeking to determine both price and quantity effects, it is
important to ensure that both the price and quantity equations are correctly
specified with all the variables affecting either term. These additional variables include sales
variables (such as lot size, average animal weight, animal sex, homogenous
lots, type of sales contract, and other characteristics that may differ between
Canadian and U.S. sales), demand shifters (such as relative prices of
substitutes including consumer income, consumer preference, demographics,
health concerns, and seasonality), and supply shifters (such as changes in
slaughter capacity in both Canada and Mexico, or decisions to export at feeder
or fed levels).
3. Including Additional Variables Is Insufficient
to Increase the Accuracy of Canada's Econometric Model
35. Even
if Canada and/or Mexico attempted to include additional explanatory independent
variables, the econometric modeling still would not provide accurate
results. Rather than focus on the actual price of
livestock, Canada and Mexico both utilize equations specified in terms of "price
basis." The flaw with this equation specification is that the
estimation of trade effects should measure how much the amended COOL measure
impacts or lowers Canadian and Mexican livestock prices. Thus, changes to the price basis, which
reflects changes in both the U.S. price and Canada or Mexico export prices, is
not appropriate because any widening basis captures both the decline in Canada
or Mexico export prices and the increase in the U.S. price.
36. Canada
states that estimating an equation "with the absolute price as the
dependent variable" will be "biased and unreliable and yield no
meaningful results that can be interpreted in the calculation of losses." Mexico suggests this approach is less
efficient and will yield a less reliable estimate than a model specified with
price basis as the dependent variable.
However, the question before the Arbitrators is not whether the "price
basis" widened or contracted due to the amended COOL measure, but rather
what quantity of livestock would be exported and at what price but for the
amended COOL measure. For these reasons,
Canada and Mexico's econometric analysis and its resulting overestimation of
the level of nullification or impairment should be rejected.
4. Canada and Mexico's Methodologies Utilize
Truncated Equations that Have Little Explanatory Power
37. Canada
and Mexico use faulty "reduced form equations" to estimate the impact
on the quantity of Canadian livestock exports to the United States and on the
price basis from the amended COOL measure.
These equations do not adequately evaluate the complex livestock and
meat industry or the relevant demand and supply shifters.
38. Requesting
parties' "reduced form equations" do not provide quantity equations
that factor in price, or price equations that factor in quantity. In particular, the price and quantity
equations, which are mutually linked (and in fact determinative), should have
the same exogenous variables.
Specifically, in a system attempting to identify both price and
quantity, two reduced form equations should be specified with price and quantity
as the dependent variables on the left hand side of the equations. On the right hand side should be all the
variables affecting the price and quantity in the livestock market. It is important for all variables affecting
either price or quantity to appear in both equations, otherwise the relevant
variables affecting price and quantity are being omitted in the reduced form
resulting in bias. Indeed, Canada itself
conceded at the hearing that its quantity equation should, but does not,
control for all causal factors. However,
Canada inconsistently – and inaccurately – does not make the same concession
for its price equation.
5. Canada
and Mexico Rely on Incomplete and Unsubstantiated Data
39. Canada
relies on unofficial weekly cattle and hog import data derived from veterinary
certificates collected by USDA's Animal and Plant Health Inspection Service ("APHIS"). This is not the appropriate data to use
because APHIS's responsibility is to ensure that health certificates are in
order, not to track import numbers for official purposes.
40. With
respect to the pricing data provided for feeder pigs, Canada notes that "no
consistent time series of price data amenable for statistical analysis is
available for feeder pigs in Canada."
Canada now seeks to rely on a limited, handpicked selection of
transactions, which are completely unverifiable. Such evidence simply cannot satisfy Canada's
burden in this regard.
41. Mexico
utilizes weekly pricing data collected by USDA's Agricultural Marketing Service
("AMS"). This data reflects a
limited sample of weekly Texas and New Mexico feeder cattle prices. The AMS price data provided is not
necessarily consistently reflective of the types of feeder cattle that are
imported from Mexico. Moreover, it is
significantly different from both the U.S. Census data and Mexico's reported
export value. The AMS reported prices
reflect both the export price and value added in the United States. However, Article 22.6 arbitration focuses on
the trade effect of the inconsistent measure.
This means it must reflect the impact of the measure on the product as
it crosses the border not any later added value.
6. Mexico's
Quantity Impact Analysis Is Also Subject to Significant Flaws
42. With
respect to evaluating the impact of the amended COOL measure on the quantity
of livestock exports from Mexico to the United States, Mexico does not conduct
an econometric analysis. Just one
omitted variable – drought – in Mexico's opinion has undermined its ability to
use econometric modeling to determine the quantity impact of the amended COOL
measure. Mexico describes at length the
difficulties associated with creating a variable to represent the economic
impact of the drought. Shifts in
producer expectations with respect to the length of the ongoing drought may
affect the timing of sales, as well as expectations about whether input prices
may be higher or less certain in the near future. Mexico notes that it is impossible to provide
a variable that would represent these unknowable and unpredictable
expectations. This alone was sufficient
for Mexico to discredit the econometric analysis of the quantity impact of the
amended COOL measure.
43. Instead, Mexico uses a simple elasticity
calculation to estimate the quantity impact.
The quantity equation is insufficient to account for the complexity of
the feeder cattle market in Mexico and the United States, much less to account
for linkages to demand for fed cattle and beef or to substitute products such
as pork. Even though Mexico's estimation only applies to one category of
livestock and level of production, Mexico's calculation should account for all
factors influencing quantity outcomes.
44. Mexico's simple calculation has two
inputs. The first is 100 percent of the price basis attributed to the
amended COOL measure as determined using the price basis econometric
equation. The United States has
explained why attributing 100 percent of the change in the price basis
estimated using this econometric technique to a change in prices received by
Mexico (or Canada) for feeder cattle (or other animals) is incorrect and
overstates the impact of the amended COOL measure.
45. The second input is Mexico's elasticity of
export supply for feeder cattle to the United States. Elasticity is a
measure of how responsive the market will be, in terms of quantity, to the
changes in price. It appears that Mexico
recognizes that a specific supply elasticity has not been previously estimated "because
of confounding effects from the drought and the COOL measure." Mexico nevertheless attempts to develop its
own elasticity. Mexico bases its
estimated elasticity on a single year, 2012, a period of time most certainly
affected by drought and other factors.
It also appears to make unsupported assumptions about the rate of
export, and ultimately with little explanation concludes that the export supply
elasticity is 4. This elasticity exceeds the appropriate level.
46. Mexico inputs the price basis estimates derived
from the econometric modeling into the calculation of export supply to
determine the quantity impact. Using a derived elasticity coupled with an
estimated price basis calculation does nothing more than compound Mexico's
methodological errors and further distance Mexico's proposed level of
nullification or impairment from the actual level of benefits nullified or
impaired by the amended COOL measure. Furthermore, using the entire price
basis estimate to determine the impact of the amended COOL measure on Mexican
feeder prices overstates the trade effect.
7. Canada and Mexico's Price and
Quantity Estimates Result in Unsupportable Levels of Nullification or
Impairment
47. Finally,
Canada and Mexico uses the inaccurately estimated quantity impact and price
basis to derive an overall level of nullification or impairment for each
livestock category. That is, Canada and
Mexico essentially multiples the price basis it attributes to the amended COOL
measure times the quantity impact it attributes to the amended COOL measure. However, Canada and Mexico's methodology
erroneously attributes to the amended COOL measure the impact of a wide variety
of other factors concurrently affecting the North American market. For this reason, the trade effect figures
provided by Canada and Mexico are unsupported and do not accurately estimate
the level of nullification and impairment resulting from the amended COOL
measure.
B. The Level of Nullification and Impairment
Should Reflect Only the Trade Effect of the Amended COOL Measure
48. Both
Methodology Papers argue to include in the level of nullification or impairment
of benefits accruing under a trade agreement estimated economic effects in
Canada or Mexico's domestic market, referred to in the Papers as "price
suppression losses." With respect
to the "price suppression losses," the requesting parties allege that
the amended COOL measure resulted in a surplus of animals in their respective
domestic markets, which ultimately "suppress[ed] the domestic price of
feeder cattle in Mexico," and "suppressed prices for livestock in Canada." There is, however, no basis under the DSU for
considering domestic price suppression as a part of the level of nullification
or impairment of benefits under the TBT Agreement or the GATT 1994.
49. First,
the DSU establishes that nullification or impairment relates to the benefits
accruing to a Member under the provisions of the covered agreements. For example, DSU Article 3.3 states that
prompt settlement of situations in which "any benefits accruing to [a
Member] … under the covered agreements are being impaired" is
essential. Similarly, Article 10.4
speaks of whether a measure already the subject of a panel proceeding "nullifies
or impairs benefits accruing to" a Member "under any covered
agreement." In this dispute,
Canada and Mexico's request to include in the level of the suspension of
concessions authorized an amount equivalent to alleged price suppression losses
is inconsistent with the DSU and goes beyond any possible nullification or
impairment of Canada and Mexico's benefits under the TBT Agreement and the GATT
1994.
50. The
request to include alleged domestic price suppression losses cannot be
reconciled with the DSU. An analysis of
the level of nullification or impairment must focus on the "benefit"
under the trade agreement allegedly nullified or
impaired "as a result of" the failure of the Member to fulfill its
obligation – i.e., as a result of the
inconsistency found by the DSB. Here, a
trade benefit under these agreements relates to international trade in
livestock, not to domestic markets.
Indeed, it is notable that neither Canada nor Mexico has, until this very arbitration, considered that the "benefits
accruing" under the WTO Agreement meant anything other than the trade in livestock.
Thus, in their GATT 1994 Article XXIII claims before the compliance
panels, Canada and Mexico claimed that the "benefits accruing" relate
to the market access of the livestock exported
to the United States, a point that the compliance panels recognized.
51. Second,
the specific DSU requirement is that the "level of suspension of
concessions . . . shall be equivalent to the level of nullification and
impairment." Even aside from the
fact that the DSU does not provide for the alleged "price suppression
losses" approach advocated by Canada and Mexico, any analysis of whether
the level of suspension of concessions is equivalent to the level of
nullification or impairment would need to account for the economic effects of
the suspension of concessions in the United States. In other words, to the extent that the level
of nullification or impairment is increased by alleged price suppression losses
to reflect broader economic effects in Canada and Mexico of the amended COOL
measure, then it would be necessary to include broader economic effects on both
sides of the equation.
52. The
corresponding level of suspension would need to be decreased by an appropriate
calculation of the broader economic effects on the U.S. economy of the
suspended trade. Otherwise, the
arbitration would not be an apples-to-apples determination of equivalency, as
required under the DSU.
53. Finally,
and again aside from the fact that Canada's and Mexico's alleged price
suppression losses are not part of the level of nullification or impairment,
Canada's and Mexico's estimates of those alleged losses are unsupported and
incorrect. Both Canada and Mexico have provided estimates that are vague, at best,
and do little to accurately assess or attribute the economic impact of the
amended COOL measure on domestic livestock transactions. For instance, there are numerous additional
factors that would need to be considered in an econometric analysis of domestic
price suppression – including Canadian and Mexican demand for livestock and
differential input costs for domestic production.
IV. Conclusion
54. For the reasons set
forth above, the United States respectfully requests that the Arbitrators find
that the levels of suspension of concessions requested by Canada and Mexico are
in excess of the appropriate levels of nullification or impairment. As described above, the more appropriate
level of nullification or impairment is approximately US$43.19 million per year for Canada, and US$49.18
million per year for Mexico, and even assuming extreme compliance costs, the level
of nullification or impairment would certainly be no more than US$128.71 million per year for Canada, and US$78.95
million per year for Mexico.
_______________