The analysis of efficiencies in superior propane: Correct criterion incorrectly applied

Full text

THE ANALYSIS OF EFFICIENCIES IN SUPERIOR PROPANE: CORRECT
CRITERION INCORRECTLY APPLIED
By: Frank Mathewson, University of Toronto and Charles River Associates and
Ralph Winter, University of Toronto and Charles River Associates
Introduction
In the Winter 1999-2000 issue (Vol. 19, No.4) of this publication, Michael Trebilcock and Ralph Winter
reviewed the state of the efficiency defence in the assessment of proposed mergers in Canada prior to the
decision of the Competition Tribunal in The Commissioner of Competition v. Superior Propane Inc.
1
In
this commentary, we discuss the analysis of efficiencies in the majority decision in Superior Propane.
We suggest that the criterion for the efficiency defence set out in Superior Propane has economic support
and adds substantial certainty to merger policy in Canada, but that the criterion was applied incorrectly in
the evidence submitted in the case.
Section 96 of the Competition Act states that the Tribunal will allow a proposed merger to proceed even if
the merger leads to a substantial lessening of competition providing the efficiency gains uniquely
attributable to the merger offset and are greater than any lessening of competition. In spite of the
treatment of efficiencies in merger cases before the Tribunal, and the publication of Merger Enforcement
Guidelines (“MEGs”) by the Competition Bureau, the interpretation of section 96 and the general role of
efficiencies in merger review in Canada were in an uncertain state prior to Superior Propane. The MEGs
(section 5.5) state that in trading off efficiencies and lessening of competition the Bureau would not
challenge a merger if the merger led to an increase in total surplus, i.e. consumer surplus plus seller
profits.
2
The MEGs’ justification for the use of the total surplus criterion, without regard to distribution
of gains between buyers and sellers, is that “[w]hen a dollar is transferred from a buyer to a seller, it
cannot be determined a priori who is more deserving, or in whose hands it has a greater value”.
3
The
Tribunal’s obiter dictum in Canada (Director of Investigation and Research) v. Hillsdown Holdings
(Canada) Ltd.
4
questioned the neutrality of transfers between buyers and sellers and left the law uncertain
as to the balancing test in the application of section 96. Howard Wetston, the Director of Investigation
and Research at the time of Hillsdown, reassured the business community that in enforcing the law, the
Bureau would continue to apply the approach adopted in the MEGs.
5
More recently, however, the
Bureau has proposed a two-stage test that departs from the simple total surplus standard and which is
quite confusing (Trebilcock and Winter, at 110-111). A senior Bureau official has also recently stated
that in cases where a merger creates a substantial lessening of competition but would pass under the total
surplus standard, the Bureau “feels that it is more appropriate for the Tribunal to determine whether the
merger increases aggregate welfare of not.”
6
The Theory of the Total Surplus Criterion in
Superior Propane
Superior Propane, if it is upheld on appeal, provides a categorical end to the uncertainty in the criteria for
balancing efficiencies and competition-lessening effects of a merger. The decision endorses essentially
without qualification the total surplus criterion. Issues of income redistribution do not matter in merger
assessment under Superior Propane: a dollar received by individuals as shareholders counts as much as a

dollar received by individuals as consumers. This endorsement of total surplus by the Tribunal was not
qualified in spite of evidence by an expert witness (Professor Peter Townley) that the redistribution of
income resulting from the merger would likely be regressive, i.e. from less wealthy individuals to more
wealthy individuals. Professor Townley had offered an approach in which the members of the Tribunal
were “invited to use their individual judgment and discretion to evaluate whether the gains to
shareholders are more or less important to society than the losses of surplus imposed on consumers by the
exercise of market power” (431). The Tribunal’s endorsement of the total surplus standard and efficiency
as an objective of merger policy is strongly phrased: “[E]fficiency was Parliament’s paramount objective
in passing the merger provisions of the Act and it intended the efficiency exception in subsection 96(1) to
be given effect. Accordingly, the Tribunal is not prepared to adopt a standard that frustrates the attainment
of that objective.” (437)
The endorsement of the total surplus standard, which stands in sharp contrast to both the Tribunal’s obiter
dictum in Hillsdown and the recent backtracking by the Bureau on the standard, is justified in Superior
Propane on the following grounds. First, in the Tribunal’s reading of the Competition Act, “distributional
concerns do not fall within the ambit of the merger provisions of the Act” (432). Second, “merger review
must be predictable. Adopting Professor Townley’s approach would result in decisions that vary from
case to case depending on the views of the sitting members of the Tribunal regarding the groups affected
by the mergers” (433). Third, the evidence showed that the transfer from buyers to sellers resulting from
the merger was much larger than the deadweight loss.
7
As a consequence “a standard that includes the
transfer as an effect under subsection 96(1) would effectively result in the unavailability of the section 96
defence” (434). Fourth, government instruments such as specific tax and other social policy measures are
more effective than merger policy as ways of meeting distributional objectives (438). Finally, the
Tribunal cites the MEGs’ support of the total surplus standard. While noting that it is not bound by these
guidelines, the Tribunal “recognizes that they contain a substantial degree of economic expertise” and
agrees with the MEGs’ observation cited above that when a dollar is transferred from a buyer to a seller, it
cannot be determined a priori who is more deserving, or in whose hands it has a greater value (439).
The purely legal arguments for total surplus as a criterion to be read directly into section 96 of the
Competition Act are less than persuasive. Just as the Superior Propane Tribunal could state “If
Parliament had intended that transfers from consumers to shareholders be considered, it would no doubt
have clearly stated this intent in the Act” (432), the Hillsdown Tribunal was able to state “If only
allocative efficiency or the deadweight loss to the Canadian economy was intended by Parliament to be
weighted in the balance then one would have thought that the section would have been drafted to
specifically so provide”.
8
Section 96 of the Act is ambiguous.
It is obviously within the mandate of the Tribunal to provide an interpretation of the statute, however, and
the economic arguments that the Tribunal draws upon to support its interpretation are persuasive.
Professor Townley’s approach, submitted in evidence on behalf of the Commissioner, invites the Tribunal
to apply welfare weights based largely or exclusively on the regressivity of the transfer from consumers to
shareholders (a “welfare-weights” approach). Are consumers of the products produced by the merging
firms among the less wealthy members of society, and if so to what extent should the transfer be regarded
as a negative outcome of the merger? In Professor Townley’s approach, the negative outcome would be
incorporated in “balancing weights” or welfare weights attached to surplus accruing to individuals of

different wealth levels. The Tribunal’s decision to reject this approach has merit. If one were to
incorporate redistributive effects in merger analysis, then it would be necessary to consider not just the
wealth of consumers but also the wealth of the shareholders of the merging firms. The income
redistributive effect of a transfer from one group of individuals to another depends on the wealth levels of
both
groups. If redistributive effects were consistently accounted for, a merger that was unacceptable
when wealthy Canadian families closely held the merging firms would suddenly become acceptable if a
teachers’ pension fund bought the shares. The incentives for share ownership by wealthy and less
wealthy individuals would then be affected, the impact of merger policy on these incentives would
become an issue for concern, and merger analysis could become hopelessly complex.
Furthermore, carrying the welfare-weights approach to its full conclusion means that mergers such as the
IntraWest acquisition of Whistler Mountain (which combined the adjacent Blackcomb and Whistler
mountain ski areas) could be accepted in spite of the prediction of negative
cost efficiencies and a positive
deadweight loss because the merger produced a favourable redistribution of wealth from very wealthy
consumers (on average) to less wealthy shareholders. A welfare-weights approach results in acceptance
of some transactions with overall negative net efficiencies, including the deadweight loss, if the
distribution of wealth is improved with the transactions. Sections 92 and 96 of the Act do not provide for
decisions involving both a lessening of competition and negative efficiencies; therefore this example
shows that these sections are not fully consistent with a welfare weights approach to merger policy.
The Application of the Total Surplus Criterion in
Superior Propane
Deadweight Loss Calculations
The application of a total surplus standard to decide whether a merger increases welfare reduces to a
comparison of the total efficiencies that are realized with a merger against the deadweight loss caused by
the merger if the merger lessens competition, causing prices to rise. From the evidence submitted in the
case, the Tribunal arrived at estimates of total efficiencies from the merger of approximately $29 million
(annualized, for a period of ten years) and a deadweight loss associated with the merger of approximately
$3 million (383, 455). The Tribunal’s estimates of the deadweight loss rest on the evidence submitted by
Professor Michael Ward on behalf of the Commissioner (455, 458).
Professor Ward provides estimates of the individual demand elasticities (or price-responsiveness) for the
merging parties in his evidence and from these simulates or predicts the price increases and deadweight
loss that would result from the merger. The average predicted price increase across subsectors of the
propane market is approximately 9% (453, 457). According to Professor Ward, prior to the merger
,
Superior and ICG each exercised market power: the estimated elasticity of demand facing the individual
firms was between -1.9 and -3.9 (or approximately -3).
9
An individual-firm demand elasticity of –3 in the pre-merger market implies that pre-merger prices are
already 50% higher than marginal cost. A profit-maximizing firm sets price according to the following
formula (called the Lerner relationship):
P – c 1
P e

where P is price, c is cost and e is the (absolute value of the) firm’s own demand elasticity.
10
An estimate
that e is 3 is ipso facto an estimate that P exceeds c by 50%.
One source of market power in the pre-merger market according to Professor Ward’s evidence is the
incorporation by each of the two major firms of the price response by its rival to changes in its own price.
Professor Ward estimates that Superior responded to price increases by ICG by matching each one
percent price increase with about a two-thirds of a percentage price increase. (Professor Ward assumes in
his evidence that the response by ICG to Superior’s price changes is symmetric.)
11
If zero market power
were exercised in the pre-merger market, in contrast to Professor Ward’s evidence, then prices would
approximate marginal cost and each firm would be a “price-taker” rather than responding to price
increases of its rivals.
Professor Ward’s deadweight loss estimates are based on Figure 1, which is produced from his
evidence
12
. In this figure, the downward-sloping demand curve represents consumers’ reservation prices
or willingness to pay for the next unit of a product. The reservation price is higher at lower quantities of
the product. Referring to the figure, Professor Ward associates a price increase from P
1
to P
2
with a
deadweight loss (“DWL”) equal to the darkly shaded triangle and a transfer from consumers to
shareholders equal to the lightly shaded rectangle. The Tribunal essentially adopts Professor Ward’s
estimates of the deadweight loss.
Figure 1 is an incorrect depiction of the deadweight loss resulting from a merger-induced price increase
from P
1
to P
2
when P
1
exceeds marginal cost (i.e. when there is pre-merger market power), a condition
recognized in Professor Ward’s evidence. Deadweight loss is measured by a triangle bounded by the
demand curve, price, and marginal cost. When P
1
exceeds marginal cost, a deadweight loss, depicted by
the triangle DWL
1
in Figure 2, already exists in the pre-merger market. That is, relative to the
hypothetical world where price equals unit cost (denoted by c in the diagram), the pre-merger market is
characterized by a loss in total surplus equal to DWL
1
. The deadweight loss associated with the merger is
the increase
in the size of the deadweight loss triangle above marginal cost as the market moves from P
1
to P
2
. This increase is depicted in Figure 2 as the difference in the two triangles bordered by the demand
curve and marginal cost, a trapezoid-shaped area. The correct merger-related deadweight loss so
calculated exceeds the area of the triangular area adopted in Professor Ward’s evidence (labeled “Ward’s
DWL” in Figure 2) by the area of the rectangle underneath this triangle (labeled “missing DWL”).
13
For a market elasticity of demand of –1.5 (the value on which the Tribunal focuses in its reading of
Professor Ward’s evidence), individual firm demand elasticities of –3, and a price increase of 9%, the
actual deadweight loss that follows from Professor Ward’s estimates is approximately 8.5 times the value
that he reports.
14
The magnitude of this error is clearly large. More importantly, the correction brings the difference
between the estimate of deadweight loss implied by the Commissioner’s expert evidence and the
Tribunal’s estimate of efficiencies well within the range of estimation error that is imposed by the
limitations of data and econometric evidence in the case. While the Tribunal notes that the Commissioner
did attempt to provide additional estimates of deadweight loss, the Commissioner did so based on
information put forward only in final argument. The Commissioner did not address this correct total

surplus test in properly submitted evidence. Because of this, the Tribunal excluded the revised estimates
(451).
As the Tribunal noted (434), citing Trebilcock and Winter, estimates of “deadweight loss triangles” tend
to be small relative to even modest efficiency gains. Deadweight loss triangles measure the surplus loss
of a given price increase when one starts from a price equal to marginal cost. As we find here, however,
the deadweight loss from a given price increase when one starts from a pre-merger price exceeding
marginal cost can be very large.
It may be helpful to elaborate on the simple economics of why pre-merger market power makes such a
difference, without reference to triangles, rectangles or trapezoids. Consider, for example, a market in
which 100 consumers each purchase one unit of the product in the pre-merger market. Suppose that price
is initially equal to a marginal cost of $10, then increases by 10% to $11. A loss in total surplus, or
deadweight loss, arises with the departure from the market of consumers whose willingness to pay is
between $10 and $11. Suppose that the size of this departing group is 10, or 10% of initial purchasers.
15
The loss in total surplus associated with each of these consumers is equal to the difference between the
consumer’s willingness to pay and the marginal cost. This loss is on average only 50 cents for each
departing consumer, for a total loss in surplus of about $5 or only 0.5% of initial revenue. The key to the
small size of the deadweight loss is that the consumers departing from the market are those whose value
for the product was only slightly above the cost of producing the product. Total gains from trade in a
market (total surplus) is the difference between the social value of the product (the aggregate of
consumers’ value for the product) and the social cost of producing the product. Where the consumers
discouraged from purchasing by a price increase are those whose value is only slightly above cost, the
loss in total surplus associated with the price increase will be small.
Now consider a second example. Marginal cost is still $10; however, the initial price is $20. The
percentage price increase is still 10% (now from $20 to $22); continue to assume that there are initially
100 consumers and that 10 of these are discouraged from buying by the price increase. In this second
example, the average departing consumer will value the product at $21. The loss in total surplus
represented by the average departing consumer is the difference in this value and marginal cost of $10,
which is $11. This is a large number. As a consequence of the initial gap between price and marginal
cost, the departing consumers are no longer consumers whose value for the product is only marginally
above the cost of production. As a result, each of the departing consumers represents the loss of
substantial gains to trade. The total loss of surplus in this second example is $100 or 5.0% of initial
revenue, 10 times the percentage loss found in the first example.
It is useful as well to address a question that may be raised in the minds of some readers. Are we
suggesting that a merger to a given post-merger price, or a merger of a given market to monopoly,
involves a small deadweight loss when the market is initially competitive but a large deadweight loss
when market power is exercised pre-merger? We may appear to be suggesting that “pre-merger market
power raises the deadweight loss associated with a merger to monopoly in a given market”. Yet this
could not be true, since the monopolization of a market that is already part way along the path towards
monopoly pricing clearly cannot involve greater social loss than the monopolization of a market with
initially competitive pricing.

The resolution to the apparent paradox is in the phrasing of the question. Monopolization of a market -
given a demand schedule and costs - involves greater deadweight loss if the market is initially competitive
than if market power is already being exercised. This proposition holds constant the demand schedule
and costs. However, the deadweight loss associated with a given price increase is larger when there is
pre-merger market power than when there is not. This proposition holds constant the magnitude of the
price increase attributed to the merger. The effect on deadweight loss of pre-merger market power, in
other words, depends on what is being held constant. In the Superior Propane merger, the price increase
of about 9% estimated by Professor Ward involves a much greater deadweight loss when its calculation
correctly incorporates Professor Ward’s own estimate of pre-merger market power. On the other hand,
relative to a competitive market the existence of pre-merger market power (assuming, for our purposes,
that it exists) reduces the deadweight loss attributable to the merger.
16
Efficiency Calculations
We move now to the other side of the efficiencies / deadweight loss comparison. The Tribunal and the
Commissioner err on this side of the comparison as well in their calculations of the cost savings
associated with the merger. Cost savings can be categorized into savings in fixed costs and savings in
variable costs. The savings in variable costs must be applied to the post-merger output quantity (Q
2
in
Figure 2), not the pre-merger quantity (Q
1
). The Tribunal adjusts downwards the respondents’ claims for
efficiencies (380) for various reasons, but these reasons do not include a reduction in volume sold as a
result of a price increase. In its trade-off analysis, however, the Tribunal assumes price increases in the
range of 7.0% to 11.0% in the various segments of the propane market (453) and adopts the
Commissioner’s assumed elasticity of demand of –1.5 (455, 458, 463). The volume decrease implied by
a demand elasticity of –1.5 and a price increase of approximately 9% is –13.5%. The Tribunal’s analysis,
in other words, assumes approximately this decrease in volume. To the extent that the respondents’
claimed cost savings (for example in categories such as reductions in the size of delivery fleet, the number
of drivers and propane supply (see 380)) depend upon volume, the claimed cost savings should have been
reduced further in both the Tribunal’s estimates and the Commissioner’s submissions.
Conclusions
Superior Propane both clarifies and relaxes the restrictions imposed by Canadian competition law on
mergers that generate efficiencies, notwithstanding the limitations on the efficiency defence of pre-merger
market power. As discussed, the efficiency defence following Hillsdown and recent statements from the
Competition Bureau rested on shaky ground. Superior Propane explicitly allows mergers that lead to
substantial lessening of competition providing the mergers generate sufficient efficiencies to meet the
total surplus test – and, as the Tribunal recognizes, relatively small cost efficiencies may be enough to
offset a substantial lessening of competition. Prior to Superior Propane, competition lawyers would have
properly advised clients not to pursue mergers that involved an obvious and substantial lessening of
competition. After Superior Propane, such advice is too conservative for mergers involving significant
efficiencies.
Will Superior Propane open the floodgates for mergers in Canada? The number of contentious mergers
brought forward may well increase, but as this paper shows the now-expanded efficiencies defence is still
limited by the correct incorporation of pre-merger market power in merger analysis. More significantly,

even a substantial increase in the percentage of contentious mergers will not represent a flood of new
mergers in Canada. In a recent period in Canada, less than 2% of mergers raised an issue under the
Competition Act.
17
The overwhelming majority of mergers appear driven by legitimate business concerns
such as efficiencies and product development, not by a desire for market power.

Figure 1: Professor Ward’s Estimation of Deadweight Loss
as Price Increases from P
1
to P
2
Q
1
Q
2
P
2
P
1
P
“DWL”
Q
Demand

Figure 2: Correct Estimation of Deadweight Loss
as Price Increases from P
1
to P
2
Missing
DWL
c
Ward’s
DWL
Q
1
Q
2
P
2
P
1
P
DWL
1
Q

Notes
1
2000 Comp. Trib. 15 [hereinafter Superior Propane. Numbers in parentheses refer to paragraph numbers in the
decision.]
2
The surplus gained by a consumer from purchasing a unit of a product is the excess of the consumer’s value of the
product over the price paid. For example, if a consumer would pay at a maximum $10 for a unit of a product, but is able
to purchase the unit at $7, the consumer earns a surplus of $3. “Consumer surplus” is the sum of such surpluses across
consumers and represents the dollar value of the consumers’ gains from trade in the market. Adding seller profits yields
the total surplus, or total gains to trade in the market.
3
MEGs, footnote 57.
4
(1992) 41 C.P.R. (3d) 289 [hereinafter Hillsdown].
5
Mr. Wetston stated that he was “of the view that, from an enforcement perspective, it is preferable not to depart at this
time from the approach adopted in the Merger Enforcement Guidelines,” Howard Wetston, “Developments and Emerging
Challenges in Canadian Competition Law,” Speech at the Fordham Corporate Law Institute, New York, October 22, 1992,
at 9.
6
Gwillym Allen, speech to conference attendees at “Meet the Competition Bureau”, Toronto, May 3, 1999, published at
http://strategis.ic.gc.ca/SSG/ct01548e.html (date accessed: February 7, 2000), at 5.
7
The “deadweight loss” associated with a price above marginal cost is the loss in total surplus in the market relative to
the case where price equals marginal cost. In other words, the deadweight loss associated with a price increase is that
portion of the additional expenditure by buyers that does not represent simply a transfer to sellers. A merger that increases
price will increase the deadweight loss in a market to the extent that demand is elastic (responsive to price).
8
Hillsdown, Section VI B.
9
Evidence of Professor Michael Ward, at 29. An individual-firm elasticity of demand of -3 means that a 1% price
increase results in a 3% decline in quantity purchased from the firm. A market
elasticity of demand of say -1.5, means that
if prices by all
firms in a market rose by 1% the market quantity purchased would fall by 1.5%.
10
Note that the relevant demand elasticity is the individual-firm elasticity of demand, not the market elasticity of demand.
11
Evidence of Professor Ward, at 26 and 27.
12
Evidence of Professor Ward, at 33.
13
The impact of pre-merger market power on deadweight loss in merger analysis is well understood in the economics
literature. For a particularly clear exposition, see Donald G. McFetridge, “Prospects for the Efficiency Defence,” 1996
Canadian Business Journal 26 (1996) at 321-357.
14
The following approximate calculations support the figure of 8.5. Refer to Figure 2. With the market demand elasticity
of -1.5, we have Q2 = Q1 [1-1.5(.09)]. The area of the triangle marked “Ward’s DWL” is then given by (1/2)(P2 – P1)(Q1
– Q2) = (1/2)(.09P1)(.135Q1) = .006 P1Q1, or 0.6% of initial sales, consistent with the value reported in Table 8 (p.34) of
Professor Ward’s evidence. The area of the rectangle underneath “Ward’s DWL” is (P1-c)(Q1-Q2) = .33P1*.135Q1 =
0.045 P1Q1. Adding the two areas yields a merger-related deadweight loss of 0.051P1Q1, or 5.1% of initial sales. This is
8.5 times 0.6% of initial sales.
15
Equivalently, suppose that the elasticity of demand is –1.0.
16
We emphasize that we are commenting on the internal logical consistency of the evidence filed on behalf of the
Commissioner and the aspects of the Tribunal’s decision that follow this evidence. In particular, we do not comment here
on the validity of Professor Ward’s estimates of pre-merger market power. Note that since the respondents did not allege
pre-merger market power, none of our comments applies to their evidence.
17
Between 1986 and 1994, the Competition Bureau examined roughly 22% of publicly reported mergers and only about
1.6% of reported mergers raised an issue under the Competition Act. (Donald G. McFetridge, “Merger Enforcement under
the Competition Act: The First Ten Years,” Review of Industrial Organization, Vol. 13, Nos. 1-2 (April 1998) 25-56. The
statistics are to the end of the 1993-1994 budget year and exclude the half year for 1986.)