HP-2020-000016 - [2025] EWHC 1451 (Ch)

When cross-examined on these points, Mr Parker was forced to concede that his revenue correlation analysis did not work on an individual brand-by-brand basis (in particular, he said, because of the different lifecycle stages of individual brands). That does, however, call into question why he provided two reports to the court which offered precisely that analysis, contending that it allowed him to identify the products that were the strongest substitutes for LOL Surprise.

Mr Parker nevertheless maintained that his revenue correlation analysis remained sound at the level of entire NPD segments, on the basis that an aggregation across products would “net out” factors affecting individual brands. But that fails to take account of the fact that even across entire NPD segments, demand will change as a result of changing consumer preferences, making such changes less informative for demand substitution. Mr Laughton explained that before the launch of LOL Surprise MGA’s business focused primarily on fashion dolls; after the launch of LOL Surprise, however, MGA’s sales of mini-dolls as a proportion of its business increased, and a number of competitors began to produce mini-dolls. By 2021, however, sales of mini-dolls had begun to decline, and Barbie was “back” and growing again as a brand. Mr Laughton concluded that “[s]uch changes in the nature of the competition happen all the time, as children’s interests alter and new trends develop”. Mr Parker’s analysis took no account of that point.

The second fundamental problem is that Mr Parker’s primary correlation analysis took as its starting point the playset dolls NPD segment, and then sought to test whether the market should be widened further by adding in the fashion dolls segment. Mr Parker thereby assumed that if he could show substitutability in general between the two market segments, then it would inexorably follow that every product in the playset dolls segment was substitutable for LOL Surprise.

That is a classic example of starting with too broad a focal set. What Mr Parker did not do was to start with the focal product, i.e. LOL Surprise, and consider which products in the playset dolls segment, or at least which types of products, were indeed substitutable for that focal product. In cross-examination, he asserted that he did not need to do so, because the logic of the SSNIP test was (in his view) that if the candidate market was tested and found to be too narrow, then it was not necessary to consider whether the candidate market was the correct starting point. That is, however, precisely the approach deprecated in the case-law and economic literature referred to at §§195–196, on the basis that it risks overlooking “pockets of market power”.

As set out above, the correct approach is to start with the focal product, defined as narrowly as possible. Mr Parker’s primary analysis did not do that. His alternative brand-by-brand analysis did purport to do that, but (as set out above) he eventually accepted that his alternative analysis was unreliable. The result was that Mr Parker did not put forward any reliable analysis which adopted the correct starting point.

Thirdly, Mr Parker’s analysis produced results which were inconsistent with any sensible assessment of competitive constraints by reference to product characteristics and functionality. The discussion at §240 above illustrates that point in relation to Mr Parker’s alternative analysis. The problem was equally evident for Mr Parker’s primary analysis, which resulted in a very broad market that included a whole range of products with wholly different characteristics and functionality to those of LOL Surprise. These included, for example:

In fact, Mr Parker’s approach did not account for the product characteristics of LOL Surprise in any meaningful way at all. MGA’s answer was to rely on Mr Harper’s comments to the effect that children wanting a particular toy would not ask for a toy based on its product characteristics, such as a “small collectible doll with a sophisticated unwrapping experience”; and that trying to define a market based on those characteristics was therefore divorced from reality.

Mr Harper was not, however, instructed as an expert on market definition and his comments did not reflect the purpose of market definition as a matter of competition law and economics. While the process of defining a product market by reference to demand substitutability will obviously look at consumer preferences and relevant drivers of demand, the definition of a market is a theoretical construct which is used to assist in the analysis of dominance. A market definition is therefore not intended to represent the actual decision-making processes of a consumer choosing to buy a particular product. Indeed, if the market in the present case were required to correspond to the way in which a small girl in the target age range would choose her toys, Mr Parker’s definition would also fail that test, since small girls do not habitually ask for toys by reference to particular NPD segments.

Ultimately, MGA had no real answer to the problem that Mr Parker’s evidence on this issue did not meaningfully reflect any assessment of the types of qualitative evidence referred to in the Market Definition Notice that are consistently used to define markets, particularly in the absence of any data on which a proper SSNIP analysis could be conducted. His approach is a good illustration of the risks of conducting a statistical analysis without verifying that its results are grounded in the market realities and common sense. The result was an opinion that gave the impression of advocacy for MGA rather than more objective evidence to assist the court.