Friday, May 03, 2019

Legal Applications of Marketing Theory, part 1

Jacob Gersen & Joel Steckel, Harvard Law & NYU Stern, Conference Introduction
Steckel gave a talk on dilution years ago and RT tore him apart (sorry!); since then, he’s done work with Chris Sprigman to answer some of the Qs I raised and I was right (not sorry!).  Has stayed interested in legal applications of marketing theory.

Saul Levmore, Chicago Law, Piece Problems: Component Valuation in Marketing and in Patent and Tort Law

Marketing literature is more interesting than legal literature here.  We have analogous problems in negligence: if 3 people jointly cause harm, how do you value the harm of each?  For patent infringement: how important to the value of the overall product is one ingredient that was found to infringe?  It’s a more interesting problem than it looks b/c of worries about over- and underdeterrence. Or a case in which someone patents playing a song on each floor of a parking garage so they can remember which floor they’re on: if they don’t increase the cost of parking after implementing this, how do you figure out what the damages are?  [He thinks costs saved in not helping people find their cars.]

Paper discusses how some solutions in some areas transfer well, and others don’t.  B/c the contribution to the joint tort car accident is a one-off, it may not transfer well. But valuation is a general problem: figuring out how valuable a third baseman is to a particular team is actually the same problem, b/c a teammate might be more or less valuable depending on who else is on the team.  The underlying relationship may not be linear.  Our task: we should look to other fields to see how they do or could solve the problem.  Though stresses that doesn’t mean the solutions elsewhere work there, or would work here.

Q: how do you think about heterogeneity of preferences?

A: in some settings we don’t care: we’re trying to evaluate a market solution. In others, he hopes it’s not a serious problem b/c mostly we can look to market solutions or ask individual people (surveying).  One possibility: ask people what it was about the product they like, or ask them to pick among products. People may not be able to name what they like, but that’s a signal it might not be important.  But he doesn’t yet know what to do with what happens if only 4 of 50 mention a feature is important to them; that might help convince a court that it’s not important. 

Q: conjoint analysis could be meaningless where other measurements of damage are better. If you have a broken fridge in a house it could knock $4000 off the value of the house—but if it can be fixed for $50 that’s what should happen instead.

A: agrees.

Q: reasonable consumer concept. Good marketers think precisely the opposite: they segment markets, then go after a segment. The whole point is the heterogeneity.  Groups do have different valuations for different aspects of products.

A: there’s some doubt, but it’s not as silly as you make it sound. General approach is to assume that law or the market is aimed at the reasonable/average consumer; if you know you’re not normal, you should speak up.  Other areas of law think that law should pick extreme solutions, forcing information out of people who are “normal”—depends on the costs and benefits of revelation.  Sometimes you might pretend to be normal to avoid the costs of valuation (e.g., you cut I pick). Usually law is unsympathetic to the non-average person, e.g., a nuisance claim by a person who can’t stand music and lives near a church.  Law is in part sensitive to the probability of fraud here.

Q: often sees calculations of actual price compared to but-for price—but what is the but-for price?  Is it a price that would’ve kept the same market share?  A price that would reflect a Nash equilibrium of a new market without the feature b/c competitors would also have reacted to the absence of the feature? A price equivalent to average consumer’s WTP for that attribute/its absence?

A: Law & econ answer: paper talks about this and he thinks there’s no single right answer.  B2B transactions would be very different from consumers.  General reaction: many ways to do it.  He wants to identify several different methods and not tell people in advance which will be used; that prevents strategic behavior. Uncertainty is a valuable part of the system though we make believe we’re committed to treating like cases alike.

Q: what happens when it’s not 6 elements but 1000, as in a cellphone? Then when you do the conjoint analysis it’s down to 6, distorting the results.

A: when it’s 1000, the chance I could build the cellphone without it is much greater, so the solution should be: if I’d known the problem upfront, what would have been my cost to avoid it? That would be the best way to go about measuring those damages—the beginning rather than the end consumer.  That might also be true when there are only 3 elements, but it’s more likely w/1000.

Peter Golder, Aaron Yeater, & Mike Schreck, Dartmouth College, Analysis Group, & Analysis Group, Assessing Trademark Strength without Surveys

Secondary meaning: in search of a more rigorous way to deal with the qualitative aspects of an inquiry. Law makes claims about how consumers perceive product designs that have implications for measurement. Secondary meaning: “in the minds of the public, the primary significance is to identify the source of the product rather than the product itself.”  Considers evidence that product features were intended to indicate source and that the firm succeeded in doing this.  It’s not just that the design needs to be recognizable but that it signal the specific purpose of identifying source, and that implies an intentionality that can be assessed.  Wal-Mart: specific assertion about consumer behavior: predisposition to equate a product design feature with a source doesn’t exist. The SCt didn’t undertake extensive empirical analysis to arrive at that conclusion, but it is the rule.

Barriers to secondary meaning: primary meaning of product design features is to provide functional or aesthetic/ornamental elements. Extent of advertising for functional benefits of features. Extent of marketplace crowding and noise with historical or current uses by competitors/third parties. Duration of existence in market.  Inconsistent use of product design features in the market.  Documentary/archival evidence can be important to these elements.  What else?

RT: I’d add: what else is in use on the product to indicate source. European concept of the limping mark that is recognized as going with, e.g., a Kit Kat, but never used to pick candy. Response: in some ways that’s about materiality.  [I agree!]

Comment: visibility/lack of visibility: consumers have easier/harder times perceiving certain things as marks.  Identifiers inside a jacket versus inside (including observability at the time of purchase, so a standard label on a jacket may be doing source identification work).  How do consumers make categories?  If you’re categorizing “birds,” hollow bones are perfectly predictive but not observable, so that’s not how ordinary people implement the category “birds.”  [I really like use of category theory though I think we’re still, as here, working out its implications.]

Standard actions to create and maintain secondary meaning: Look for internal planning documents about intent to make features source identifying; look for documents w/clear communication objectives for attempting to establish secondary meaning and measures progress; company carries out planned communication (e.g., look for ads); advertising “famous,” “iconic,” “exclusive,” “unique,” “signature”; aggressively policing asserted marks to protect exclusivity.  Qs? [Also interestingly, this framework would be consistent with Mark Lemley’s argument that having—and thus allowing/not policing against—parodies should be a requirement for fame, which would turn into media coverage rather than competition.]

RT: This is a legal question: the doctrine right now requires none of this, because the theory is always if there is consumer confusion then there is something protectable.  So you can have all this and it won’t cover the waterfront unless courts also say that its absence is dispositive.  There are cases/proceedings finding protection without any of these things, e.g., the TTAB saying that the University of Wisconsin can fail to police/“impliedly license” its marks for 70 years and then (re)claim its TM rights.

Comment: but perhaps it’s just implausible that this creation of secondary meaning will really happen without the marketers picking up on it and talking about it internally so there will be documentation.

Marketplace outcomes related to secondary meaning: Complement to surveys: media provide evidence of success or failure of calling out features as source identifying. Traditional/social media.  Company websites (including past versions). Online search behavior—if they are source identifiers, consumers should be trying to search by them at least to some degree [which raises the limping mark issue again]. Online reviews. % of sales w/product design features. Company monitors and documents progress towards achieving communication objectives related to creating or maintaining secondary meaning.  [I like the suggestion, which I’m not sure has shown up in the cases, that advertising functional features has a separate impact on secondary meaning—it’s not just that it indicates functionality; it also tells consumers that they shouldn’t rely on the touted feature to indicate source.]

The law on what’s required for secondary meaning varies by circuit.  [They categorized intentional copying by D as an “outcome” but that doesn’t fit in their framework at all. The missing concept: functionality—if there are good noninfringing reasons to copy, then copying tells us nothing about secondary meaning. That factor shouldn’t be in the legal tests, at least without requiring intentional efforts to confuse in particular and not just to copy, and their framework helps explain why.]  If you can luck into secondary meaning, why would courts care about “look for” advertising?

Cass Sunstein, Harvard Law School, Popcorn: Mandatory Information Disclosure:
How to value the benefits of information? Pervasive unmet challenge in policy all over.  Principal focus is on regulatory agencies, but courts and private sector entities are also trying to value info.  Toy/discussion framework: people might want information for instrumental value—they can know whether to buy a product, get health care, change their lives. Our primary approach in the past. There’s also hedonic value: information might make you happy or sad, and people might be willing to pay to get information that makes them happy, or willing to pay to avoid information that makes them sad. Cognitive value: learning may be something that people value. Maybe it’s just curiosity—how far from the earth to the moon? Or maybe it will reinforce their model of the world, or maybe it’s intriguing if it undermines their model of the world.  Sign issue: the valence of instrumental, hedonic, cognitive value may be positive or negative.  Learning your client is guilty may have negative value. 

Recent data: knowledge is not always preferred. Mesolimbic reward circuitry selectively treats the opportunity to gain knowledge about favorable but not favorable outcomes as a reward to be approached. WTP to receive or avoid knowledge was tied to participants’ expectations about whether info would be positive or negative.  Roughly 1/3 in trials chose ignorance.

Asking consumers whether they’d want to know: if they’ll get Alzheimer’s, whether their partner cheats, year of death, number of calories, whether there’s heaven (slightly lower percent wants to know whether there’s hell), predisposition to get cancer.  Alzheimer’s: 47% want to know, $107 WTP (average; median is significantly lower). Cancer: $115, 58%.  Spouse cheats: 57%, $121; Death: 27%, $154; calories, 43%, $49 (annually, contingent on wanting to know); weekly cost of appliance operation: 60%, $44.  Safety ratings of tires: 67%, GMO, 60% $101, conflict minerals 55, $109, Online performance of airlines, 57%, $105, GHG emissions from car, 57%, $110. Some would pay not to receive calorie info, apparently for hedonic reasons.

On yes/no, a lot of heterogeneity out there. Usually around 55-65% even want to know, for typical disclosures. Two categories of not wanting to know: sometimes it’s bad news, and many people don’t want bad news; sometimes it is who cares. WTP numbers vary and are usually pretty low.

There are a lot of labels out there: calorie, fuel economy, energy efficiency, conflict minerals, graphic warnings, country of origin, greenhouse gass, nutrition, dolphin-safe tuna.

Doing CBA: Four approaches. (1) Benefits not feasibly quantifiable, so silence is golden (fuel economy labels, conflict minerals).  Common but he hates it.  Q re actual behavior v. expressed WTP: there’s limited evidence about this, but people who use calorie labels are likely to have high self-control whereas people who don’t want calorie labels don’t have high-self control.

Observation: WTP of those who want to know when they’ll die is higher even though the percentage is low—preference intensity varies.  Payment willingness might be about how likely it is you’ll get the information some other way.

Comment: hedonic value of information may be one-time whereas cognitive is persistent. 

A: Hedonic damage of knowing the year you will die might be long-term.  More generally, the toy model here uses a rational actor model, but that’s not complete.  People don’t want exposure to info inconsistent with their political beliefs. But on average people are mistaken in hedonic forecasting: they aren’t as unhappy with getting inconsistent info as they thought they’d be, in intensity and duration. Might be present bias or failure to forecast adaptation. 

Q: why ask such self-directed questions? A lot of policy Qs will be: do you want to know how many people will die in our next war?  You might not be WTP to know the capital of New Zealand, but be WTP for everyone in the country to know that.

A: our labeling Qs are asking about whether you benefit from a label.

Q: but if you asked “do you want your kids to know the calories in their food,” the answer might be very different.

A: is that the right Q for info disclosure benefits?

Q: better than saying “do you want to know.”

A: you might want to know the benefits your children get from that.

Q: research on organizations suggests organizations may have special difficulties processing information they’re not set up to receive.

Some discussion about the “year you’ll die” question: did they believe it? Did they think they could “fight” it once they knew? Sunstein thinks the survey had an implicit “work with me here” message and that people got it.  In Europe, Diet Pepsi is Pepsi Max—Diet Pepsi was punitive, “good for me.”  That maps onto a lot of an actual driver of responses.

He thinks the data quality of MTurk varies from excellent to pretty good—if you compare results to nationally representative surveys, they’re usually pretty close—he wouldn’t expect a huge variation though some percentage variation wouldn’t be surprising.

Commenters: one had really bad MTurk experience; another has had varying experiences. Depends on the kinds of Q you ask; these Qs were interesting and meaningful and some marketing Qs would be less so.  Some evidence that MTurkers in some parts of the world sit in the same rooms and consult on answers.

Q about relationship of answers to how optimistic people are overall: e.g., date of your child’s death (he thinks learning it would always be painful, though I don’t know—2115 might be a pretty impressive answer, albeit an unlikely one).  Optimism bias, hedonic forecasting error, illusion of control—distorting factors. Magnitude of effect depends on Q.

For policy analysis, we’re trying to look at end states: e.g., for energy efficiency labels, how much would society gain in reduced emissions/particulate matter. That doesn’t capture all of the relevant values if people are made sad by the information.  Got interested in that when calorie labeling was extended to theaters and he got the response “you ruined popcorn!”  Marketers intuitively understand this—but there’s less understanding of the relationship b/t instrumental and hedonic or of the malleability of the hedonic.

Q: relationship between avoidability of info/display and preferences?  E.g., calorie labels that are impossible to miss [nothing is impossible to miss] versus on the back.  Home energy reports: a lot of people don’t like them even if they will save money with them.  Might suggest that you could respond to heterogeneity by differing presentation/availability.

No comments: