Good morning and welcome to the fourth paper presentation in the Second Annual Conglomerate Junior Scholars Workshop. Todays' paper is A Coasean Analysis of Marketing by Eric Goldman. The expert commentary, provided by Peter Huang and Frank Pasquale, will appear in separate posts under this one during the day. We invite all readers to comment on Eric's paper in the commennts section of this post. I have heard Eric present this paper, which espouses a contrarian view of "spam," at two conferences, and I look forward to the discussion here.
The Glom views this workshop as an academic conference in cyberspace. Therefore, any anonymous comments will be deleted. As with a conference environment, if you have comments you wish to share with Eric privately, feel free to email him with those comments. The abstract of the paper is here:
Consumers claim to hate marketing—mostly, because they get too much unwanted marketing. In response, regulators develop medium-by-medium marketing suppression regulations. Unfortunately, these ad hoc solutions do little to satisfy consumers, and dynamic technologies and business practices quickly render them moot. Instead of continuing this cycle, there would be some benefit to developing a cross-media marketing regulatory scheme. However, any holistic solution must be predicated on a clear rationale for regulating marketing. The most common justification is that marketing imposes a negative externality on consumers, but this argument ignores the private and social welfare created by marketing and can lead to cost overinternalization and marketing undersupply. The Coase Theorem also suggests that social welfare improves by reducing the costs of matching marketers with interested consumers. To achieve this, consumers need a low cost but accurate mechanism to manifest their preferences. This Article shows that typical regulatory and marketplace solutions do not provide effective mechanisms. Instead, marketer-consumer matchmaking will improve from technology that will automatically infer consumer preferences and use these inferences to filter incoming marketing and seek out wanted content. This technology does not yet exist, but it is being rapidly developed. However, regulation of surreptitious monitoring devices (like adware and spyware) may inadvertently block the development of this socially-beneficial technology. As a result, current regulatory overreactions to developing technology may counterproductively foreclose social welfare improvements.
TrackBack URL for this entry:
Links to weblogs that reference Junior Scholars Workshop: Eric Goldman, A Coasean Analysis of Marketing:
1. Posted by Jeff Lipshaw on July 5, 2006 @ 8:56 | Permalink
On quick read, an interesting idea, and engagingly written, but one affected by the hammer-nail problem (when you have a hammer, every problem looks like a nail).
In the end, the idea that technology, as a form of market alternative, will solve the problem, does seem to me to be the right answer, and it may even be that the Coaseian analysis is right. In a nutshell, the problem has to be the relative low cost of the marketer in mass-mailing a message from Splendiferous B. Dogood selling Viagra, regardless of the infinitesimally small cost to the consumer in deleting it.
But I think the paper would be better off by digging deeper merely than the "Do Not Call" registry for analogs to the equivalent of spam, and how marketers and consumers react to it. I receive all sorts of unwanted information every day, from billboards, to junk snail mail, to television commercials (how many times will Chrysler run the "Ask Dr. Z" before I come to hate Dr. Z?). I think this strengthens Eric's argument by the way - we seem to have come to terms with these old technology intrusions, and chances are we will find ways to come to terms with the new ones as well. (As an example, I wonder what happened to the viewership of American Movie Classics versus Turner Classic Movies once AMC decided to run commercials. I, for one, stopped watching it.)
The paper touches on but slides by another issue that is worthy of some comment - the extent to which we are shaped by media (the Frankfurt school criticisms) versus freedom and autonomy, but I think that's again part of power of the economic hammer. Whether each little monad out there is fully rational or boundedly rational, we jump to solutions informed by economic means before we consider the implicit ends. If you reside on the critical end of the philosophical/political spectrum, and think that the media and information revolution has fundamentally changed something about human society and culture, you might well find yourself on the consumer entitlement end of the regulatory spectrum. And if the essence of freedom and personal autonomy run pure in you ("if you don't like what you are seeing, turn the damn thing off"), marketer entitlement seems like no skin off anyone's nose.
Thanks for using the power of this medium to an interesting end!
2. Posted by -dsr- on July 5, 2006 @ 13:39 | Permalink
I'm afraid you are overly optimistic when you write Instead, marketer-consumer matchmaking will improve from technology that will automatically infer consumer preferences and use these inferences to filter incoming marketing and seek out wanted content. This technology does not yet exist, but it is being rapidly developed.
No matter how cheap the technology, it will still have a non-zero cost. There is a natural race to the bottom among mass-marketers in providing the widest possible coverage; thus, the delta between "preference-matched" and "indiscriminate" messaging will ensure that indiscriminate marketers continue to thrive.
Only a significant cost increase or similar negative externality can overcome that delta. In the case of the Do-Not-Call List, the negative externality was provided by the legal punishment asserted against those who failed to check the list. In the ongoing turbulent jurisdictional dispute of the Net, this is unlikely to have much effect. (Also note the large number of spammers who employ illegal or legally dubious means anyway... there will be little deterrent effect.)
3. Posted by Eric Goldman on July 5, 2006 @ 15:46 | Permalink
Jeff, thanks for the comments. There has been a lot of attention paid to the supply side of marketing (basically, how to increase marketer costs to reduce supply) but comparatively little attention paid to the demand side of marketing. Ultimately, more than anything, my paper is about empowering consumers to control their information intake. We cede lots of power to government and media intermediaries to control our information flows, and technology could enable us to take back some control without experiencing information overload.
DSR, thanks for the comments. You remain focused on the supply side. Consider the arguments from the demand side. A particular marketing item, even from "indiscriminate" marketers, can have net positive social value. So before we throw out those messages through regulatory intervention/cost shifting, let's look for ways to get the message into the hands of those who value it and keep it away from those who don't. Ultimately, client-side technology can do this without the adverse consequences of cost overinternalization.
You also raise the important limits of legal enforcement. Our experience with the Internet has shown that many actors simply ignore prevailing law, which moots any implicit cost shifting from regulation. Again, client-side technology makes those lawless decisions less important.
4. Posted by Paul Gowder on July 5, 2006 @ 15:48 | Permalink
I haven't read the paper yet -- I'll do so tonight and maybe be able to chime in with more substantive comments on this interesting topic. However, from reading the official commentary here and the abstract, let me toss out one thing here, tangentially related to Peter Huang's comments. ...
To what extent is preference-matching marketing even worse than non-preference-matching marketing? Consumers who have poor self-control, or who are unusually susceptible to advertising, might be worse off if they receive advertisements for products they actually want, because it will lead them to spend an inordinate amount of money.
To be concerned about this problem requires tossing out a little bit of revealed preferences dogma (surely no major loss) and a little walk into the behavioral law and economics wild side (particularly see the most recent series of Jon Hanson "critical realism" articles). Intuitively, it seems like these are reasonable steps to take! After all, if I'm addicted to candy bars but not to porn, surely I'm worse off if I receive more candy bar advertisements and fewer porn advertisements, as I'm led to waste more money.
5. Posted by Eric Goldman on July 5, 2006 @ 16:56 | Permalink
Paul, good Q. I could state it differently--is targeted marketing worse than untargeted marketing? I would staunchly argue no. Untargeted marketing has a much lower likelihood of being useful to any individual recipient. Therefore, if anything, I would argue that social welfare would improve if we *required* marketers to target their marketing.
However, there is some not-insubstantial support for the argument that untargeted marketing is better. For example, consider the lawyer advertising cases, where the Supreme Court has put much tougher restrictions on targeted marketing than untargeted marketing. So one might argue that, at least to the Supreme Court, untargeted lawyer marketing is preferable in some circumstances to targeted lawyer marketing for precisely the reasons you identify.
As another argument against targeted/preference-based marketing, see Tal Zarsky's chapter on online persuasion at http://law.haifa.ac.il/faculty/lec_papers/zarsky/OPTaP.pdf .
I'll have more to say on this in response to Frank's discussion of preferences as exogenous vs. endogenous to marketing.
6. Posted by Eric Goldman on July 5, 2006 @ 17:41 | Permalink
This comment responds to Peter Huang's post. Peter makes a suite of excellent comments, and I'm grateful for his input and expertise. For efficiency, I'm not going to do a point-by-point, but there are a few conceptual comments that are particularly interesting/thorny.
Peter questions the empirical support for media convergence and cross-media advertising elasticities. With respect to convergence, we may disagree about specifics and timing, but convergence seems inevitable. Bits are bits, after all. Personally, I think it's already impossible to define media technologies with legal precision, but perhaps that reflects my own drafting skill limitations.
Regarding elasticities, Peter is right that I don't have any empirical data quantifying the elasticities. If anyone has such data, I would gratefully welcome it. However, based on my anecdotal experience at Epinions, I found that our advertisers simply wanted qualified customer leads. They didn't really care where those leads came from. If the leads could be obtained more cheaply from other media, the advertisers would shift their business in that direction. So, at Epinions, we weren't "really" competing with other online sources of traffic per se; instead, we were competing with the complete panoply of traffic-generating sources. And if we weren't delivering cost-effective leads, we weren't going to keep the advertisers.
I will need some time to digest Peter's comments about latent preferences. I refer to latent preferences as those consumer wants/desires that we have but can't articulate. Impulse purchases are an example. We might not have an active search for such items, but they still generate positive utility.
Another example is a product that fills a need that I have but I don't know the product exists. There's no way for me to articulate the preference in any meaningful sense without knowing about the product; so exposure to that product coalesces the desire into an articulatable want.
Perhaps the term "preference" is creating some of the problem. There's no hierarchy between "active" and latent preferences; each generates some utility for the consumer, but only the consumer knows how much (and perhaps can't even articulate/quantify this). So maybe the terms "consumer want" or "consumer desire" would make this clearer.
I think Peter's analogy to patients' conversations with doctors is very interesting. I haven't gathered all of the literature, but I've been able to find a fair amount of evidence that consumers misreport their preferences. For the latest article on the topic, see Jason Fry, Under Recommendation Engines' Hood, WSJ, http://online.wsj.com/public/article/SB114961753581872822-jcRlO7GbCpnPIHxI4kKqWTOufgA_20070611.html . So my goal is to stop asking consumers questions where they may not accurately report their preferences. Instead, Coasean filtering technology watches consumer *behavior* and see what the consumer REALLY thinks (as manifested through action) rather than relying on self-reported preferences. Accordingly, the technology can overcome the problem that Peter refers to by its very operation. (Unfortunately, I can't think of an analogous technology for the medical situation--maybe a tricorder from Star Trek?).
I gratefully acknowledge the various technical suggestions and pointers to other literature. Peter, thanks for taking the time to read the paper and provide such thoughtful comments.
7. Posted by Eric Goldman on July 5, 2006 @ 18:13 | Permalink
This comment responds to Frank Pasquale's comments. Amidst Frank's kind words (which I appreciate very much), Frank raises some fairly devastating critiques of the paper that I don't have any good responses to.
First, he's right that some marketing creates consumer preferences, while other marketing merely informs consumers how to fulfill an existing preference. I cannot offer any way of distinguishing between the two. In part, I'm agnostic about this issue, because I do not inherently object to preference-creating marketing. My attitude is that if it creates private utility, then I'm not going to question the validity of that utility. However, I have no rigorous way of defending this position against the critique that manufactured utility may be less meritorious/socially beneficial than exogenous utility.
Second, I also have no adequate response to the relative merits of a counterfactual world where marketing simply didn't exist. At minimum, I can't prove that social welfare is greater in a world with marketing than a world without marketing. Instead, I made some (unstated) assumptions that consumers have desires for goods and services; and that so long as it's profitable, marketing will exist to cater to those desires. However, I can't defend the proposition that this world maximizes utility when the various costs of marketing are fully considered.
I also assumed that there would be no realistic way of outlawing all marketing. Such a law could be written (let's ignore all constitutional considerations here), but marketing can be manifested in so many ways that I think marketing will occur so long as people can communicate, even if we had a law banning *marketing.*
In my (limited) defense to Frank's meta-critiques, I architected this project from a fairly microeconomic perspective--my goal was to try to resolve the dilemmas posed by marketer-consumer communications, and in particular respond to the various commentators who viewed these communications as an externality. Doing so allowed me to sidestep Frank's broader contextual/macroeconomic concerns, for better and for worse.
I'm very grateful for the thoughtful remarks and various pointers provided by Frank. Eric.
8. Posted by Paul Gowder on July 5, 2006 @ 20:04 | Permalink
Oh, gosh, how embarassing, I missed that part in the middle of Frank's response. What he said! Although I don't think he goes far enough -- it's not just preference creation we should be concerned about, but oversatisfaction of existing preferences. If I am induced to satisfy my preference for double quarter pounders with cheese too many times, my utility function goes way down even though McDonald's didn't "create" those preferences. (But again, see Hanson, and on this one, see his Obesity and Justice piece.)
As promised, some additional thoughts. This comment is very lengthy, but I hope it is helpful. (I'm trying to singlehandedly refute Kate Litvak's "Bugged Water Cooler" thesis I guess.) Overall, the article is very interesting and seems to make a great contibution to the debate on advertisement and "coasian filters."
-- It seems to me that the initial premise that regulatory interventions are poorly conducted is all but unargued. The paper asserts that "regulators make the same systematic regulatory errors with each new medium," but that seems to be unsupported (especially in light of the previous observation that these regulations are created on a case-by-case basis, which would suggest that systematic regulatory errors would be less likely). Nor is it convincing that the development of technology will break medium-specific regulation. Consider, for example, cigarette advertising on television. Development in the television technology over many years has not made that regulation obsolete.
Later, the criticism of specific forms of regulation is limited largely to their less than ideal effectiveness. For example, opt-in regulation is criticized in part because it is medium specific and because consumers can be tricked into opting in, and thus it is only partially successful in stifling unwanted messages. However, a partial solution is better than no solution at all. (A small point on this: the dismissal of the mandatory metadata option ignores the possibility of automated sorting. Filtering software that would filter out finely granulated metadata from e-mail would be trivial to design with sufficient data requirements. Heck, if you make porn advertisers put [[porn]] in their subject lines, Eudora can do it out of the box. That would greatly reduce the cost to consumers.)
Of course, this flaw is not fatal to the argument. If marketing is good, regulation that reduces the net amount of marketing is bad whether that regulation is itself poorly designed or not. However, the attempt was made to suggest that regulation of marketing is doomed to failure, and it doesn't seem to fly.
-- More significantly, the attempt to avoid negative externality and tragedy of the commons arguments appears deficient. The answer offered to those who point out the negative ACU and probably negative RU is, in the first instance, that these may be outweighted by a positive SU. True enough. However, one does not justify negative externalities by pointing to the potential of a net positive utility in each individual case. (The argument that electric power is a social good, for example, does not justify failure to internalize pollution costs.) The reason for this is obvious -- a seller who fails to internalize some costs is per se engaging in an inefficient level of production by producing more advertising than is justified by the expected revenue. That is why all costs must be internalized, so it is correct to point to any externality as a social harm from advertising, even in the face of a possible net benefit to any individual advertising consumer.
Perhaps a better analogy is to a loud rock band. If I own a nightclub that plays loud rock music through thin walls, some of the neighbors might like it. Nonetheless, it would be more efficient if I were made to internalize the costs to the neighbors who didn't like it, since that would lead to my accurately calculating the costs and benefits of a given act at a given volume.
I'm not sure if the Van Zandt, Kraut, and Loder references you give in fn 72 refute this point. Intuitively, however, it would seem unlikely. The notion that marketeers fail to internalize the benefit to consumers from marketing (which is your description of the Van Zazndt ref.) is implausible on its face -- the benefit to a consumer from beneficial marketing is that they learn of a product they want to buy. This is exactly the benefit that marketing is designed to capture, measured in probability of making a sale.)
-- In general, I think the repeated claim that marketeers forced to internalize one cost will just turn to a medium where they are not forced to internalize costs (particularly as stated in the discussion of attention marketplaces) must fail. What are these media? Established media already make marketeers internalize the costs of their advertising. Marketeers must pay to get on television, and thay payment represents, in part, the negative utility caused to consumers by advertising (expressed by the fact that the television companies have to pay the consumers with entertaining content to get them to watch it). The same can said of magazines, newspapers -- several of the most prominent and effective forms of modern advertising are indirectly based on exactly the "attention marketplaces" that are dismissed as a failed feature of the .com bubble.
-- The discussion of coasian filters in the deliberative democracy content is misleading. First, it's not just deliberative theories of democracy but all theories of democracy that would have reason to be concerned about filtering of political messages. Deliberative or otherwise, anyone concerned about democracy should be concerned if someone with known Democratic Party preferences is prevented from seeing advertisements by Republican Party candidates. Also, many of the claims in this section are unsupported. For example, who says that consumers just ignore or avoid content that generates negative NPU, or that serendipitiously exposed political advertising only helps the democratic process when consumers are forced to pay attention? Isn't it possibly the case that my clicking past a Republican advertisement that briefly flits on the screen might, if it is well-designed with an attention-grabbing message that appeals, will cause me to give momentary consideration to their ideas?
Also, political advertising seems (this is impressionistic, mind) to be vastly less prevalent than commercial advertising. Refusing to permit coasian filters to be applied to political advertising would hardly be likely to create such a glut that people would "leave the medium entirely."
The claim that consumers is suffering from information overload is -- uh -- dubious. The citations for this claim (fn 299) are questionable. Shenk's book is more like propaganda than anything else, and, as I dimly recall from reviewing it a few years ago, doesn't actually cite any real evidence for this "overload." Especially -- especially -- as to political communication. The other source seems troubling just from a quick google.
The "daily us" claim is more interesting. It may be possible that some sort of coasian filter would be able to track latent political preferences as well as latent marketing preferences. However, I think the goal of political communication is different from the goal of marketing communication. Preference creation (persuasion) is an acceptable goal of political communication, whereas it's at least kind of shady in marketing communication. It follows that some political communication might best reach more than just those with latent preferences. Also, while some latent preferences might be obvious targets for political advertisements and easy to track (someone who vists church websites might be an easy target for biblical appeals on any number of issues they may not have considered), others are tougher. It's hard to imagine the sorts of activity that might indicate a latent preference for affirmative action, or for the end to a war.
The claim that marketing educates consumers about the marketplace, and thus gives them information necessary to guide the government in regulating it, is nothing but speculation. Marketing hardly provides the kind of information necessary to make objective political judgments, and hyper-targeted marketing is more likely, rather than less likely, to provide heavily biased information designed to play on consumer's individual rationality.
Some smaller points...
-- A description of the business model of the "infomediaries" might be interesting, incidentally. It would help readers evaluate the claim that they failed for some relevant reason.
-- Footnote 224 contradicts the general thesis of the paper. The rational consumer would not seize the bond in all cases because the rational consumer would, on the assumptions motivating the paper, have an incentive to not deter a certain level of advertising (e.g. the most beneficial advertising). A consumer who gets an ad for one good product from a company and has reason to believe that she might learn of more good products would not rationally punish the advertising. At worst, there's a collective action issue there related to the deterrent impact of each bond seizure.
9. Posted by Frank on July 5, 2006 @ 21:37 | Permalink
Just two little thoughts:
1) In response to Eric: I don't think I'd like to see a world with no marketing. Just an America where there is less marketing, or where marketing were less effective. I admit that this is a more likely than not an unpopular normative judgment based on an aversion to our generally low savings rate and conspicuous consumption/waste.
I do agree that you've framed the scope of the paper in a way that allows you to sidestep most of the larger themes I've introduced. I hope that projects like this one can eventually be made more receptive to concerns about the "good society" and "the good life." But I also realize that rigor often only comes with focus.
2) To Paul: just a little point about info-overload--I'd love to see your review of the Shenk book. I really liked Shenk's work, and I have a piece on SSRN on "information overload externalities." I do think they are real, and I think Goldman is right to be looking at ways of helping consumers sort through a deluge of data.
10. Posted by Peter H. Huang on July 6, 2006 @ 10:42 | Permalink
Just a quick response to Eric's, Paul's, & Frank's posts:
Many neoclassically trained economists assume that rationality of people's behavior means that economists caninfer an individual’s private, subjective, and unobservable preferences from that individual’s public, objective, and observable behavior in terms of their actual choices. This revealed preference approach requires that preference orderings are well-behaved, not only in the sense of satisfying certain technical axioms, such as the weak axiom of revealed preference; but also, more crucially in the sense of being stable, both across contexts and over time. Much of recent behavioral and experimental economics research empirically demonstrates how much so-called preferences are constructed from, and sensitive to situational contexts or influences. In particular, for many things, people do not have preferences that are fixed and given exogenously. Instead preferences are inchoate and formed endogenously, in part in response to marketing. This changing of preferences raises thorny issues for measusing individual and social well-being. If people have adaptived preferences, then should legal and policy makers utilize their pre- or post- marketing preferences to measure their well-being? This multiplicity of preferences also raises issues about meta-preferences, e.g. "I'd like to be someone who doesn't like chocolate or succumb to Hershey ads."