In April 2015, the Quebec Court of Appeal released its decision in Dunkin’ Brands Canada Ltd. c. Bertico inc. 2015 QCCA 624. The Court of Appeal upheld the trial judge’s ruling (Bertico inc. c. Dunkin’ Brands Canada Ltd., 2012 QCCS 2809) that Dunkin’ Donuts did not make appropriate efforts to stave off the impact of the rapid expansion of the rival Tim Horton’s chain into Quebec, and agreed that Dunkin’ Donuts was liable for having breached its contractual duties, as well as its duty of good faith to its Quebec franchisees. Where the Court of Appeal did find take issue was with some elements of the trial judge’s damages award, reducing it from $16.4M to $10.9M. In the next two posts, I will comment on some of the interesting damages of the aspects of the appeal. This post deals with the issue of the proper application of the “benchmark” approach to lost profits calculations.
Dunkin’ Donuts had historically had a strong brand in the province of Quebec. Beginning in the mid-1990s, Tim Horton’s made a strong play to expand into the Quebec market, at the expense of Dunkin’ Donuts. In spite of repeated pleas from franchisees for assistance from the franchisor, Dunkin’ Donuts share of the Quebec market declined precipitously, from 12.5% in 1995 to only 4.6% in 2003, and many of its franchisees were forced to cease operating altogether due to unprofitability. A group of 21 plaintiff franchisees were ultimately awarded damages for both a) lost profits and b) loss of investment/business value.
The Damages Award
At trial, the parties put forth competing models to project the sales at the plaintiff franchise locations “but for” the failure of Dunkin’ Donuts to properly support the system. One model proposed by the plaintiff’s expert accountant was to assume that the plaintiff stores would have grown their sales at a rate of growth equal to that achieved by Tim Horton’s during the damages period. This methodology is generally referred to as the “benchmark” approach (it is referred to the “comparable” approach by the trial judge ). Under this methodology, one projects a plaintiff’s sales during a loss-affected period, but for the alleged wrongdoing, based on the actual results of a similar yet unaffected firm during the same period.
On appeal, Dunkin’ Donuts noted that Tim Horton’s primary competitor in the period in question was, in fact, Dunkin’ Donuts. Using Tim Horton’s actual growth rate during the loss period was not an appropriate measure, since Tim Horton’s actual rate of sales growth in Quebec had been positively affected by the failure of Dunkin’ Donuts to support its brand. In projecting store sales in the hypothetical “but for” world, in which Dunkin’ Donuts is assumed to have acted properly, Tim Horton’s sales would have been lower, and it is this lower growth rate that ought to have been applied. The Court of Appeal accepted this argument, and decided that it would be more appropriate to use 75% of Tim Horton’s growth rate to project the plaintiffs’ sales.
Problems with the Benchmark Approach in Franchise Litigation
The (mis)application of the benchmark approach in the Dunkin’ Donuts case illustrates both the attractiveness of the benchmark approach, as well as one of its drawbacks.
The benchmark approach to projecting sales can, at first glance, be very appealing, and especially so in franchise litigation involving an individual franchise, where there would appear to be many businesses that are very similar to the plaintiff, whose results can be used as a sort of “control group”.
The danger is that one might select a control group that is so similar to the plaintiff business that the damage caused to the plaintiff actually affects the control group as well. To illustrate, consider the following example.
A number of years ago, I had to quantify lost profits at a particular location for one of the largest franchise restaurant systems in Canada. The location (the “Loss Location”) had experienced flood damage caused by one of its contractors, and had to close for approximately 8 months (the “Loss Period”). I considered two potential approaches to projecting lost sales :
- Based on the rate of year-over-year growth experienced in the 12 months prior to the incident.
- Based on the actual rate of sales growth of nearby, unaffected stores during the Loss Period.
Both of these approaches proved problematic. Year-over-year growth at the Loss Location had been spectacularly high; upon inquiring, I was informed that nearby locations of that chain had undergone renovations the prior year, causing a diversion of customers to the Loss Location. Such a spike in sales would not have persisted following the completion of renovations to the nearby stores.
And of course, results at nearby stores during the loss period were skewed by two factors: the diversion of customers who normally would have dined at the Loss Location to other, nearby locations; and the natural increase in sales at the nearby locations following the completion of their renovations.
The correct approach in that case (or at least, the one I adopted) was to:
- Use a pre-incident base period that eliminated the effect of the one-time increase in sales at the Loss Location due to renovations at nearby stores; and,
- Use a broader benchmark group, consisting of all stores outside of a particular radius of the Loss Location. These stores would not have received much in the way of diverted customers, and the sample group was large enough that the effect of location-specific events such as renovations (which were being carried out on a staggered basis within the system) would be only minimal.
Reconstructing the “But For” World
The issue dealt with the Quebec Court of Appeal also suggests a broader frame of analysis.
It is tempting to assume that the effect of a wrongdoing has been limited to the plaintiff, and that otherwise results in the actual world were unaffected and can be used to recreate the “but for” world. But, perhaps more often than we care to believe, a wrongdoing involving a single company can have a sort of “butterfly effect” on things such as market size price. This can be the case even when the injured party is only a small part of its market. I will provide other examples of this in future posts, but for now I conclude with the following lesson:
Lesson: If using an outside benchmark to project the results of the plaintiff but for the wrongdoing, be sure to fully consider the possible effect of the wrongdoing on that external benchmark.