Economics of Digital Advertising

Posted on June 10th, 2018 by Hoagy

We are in a strange economic situation where huge swathes of our economy function not by charging directly for their services but by selling advertising for products in other sectors. Who are these people? Google and Facebook have a dominant position, but what is it about these industries that allow them to use this business model? As we know, there is a value to data, and this data is sold in all kinds of ways.

Why is a new paradigm needed to answer this? Have products created by their ability to sell adverts been around before? The free TV model is the most significant precursor to this free w/ ads model. Two big differences stand out:

1: The personalisation of the data that is collected. This has been in the news recently due its deleterious effects on privacy but its most economically relevant factor is the way that it encourages monopoly, because, assuming no large differences in the intelligence applied to the gathered data, the company with the most data will provide the most advertising revenue, and the larger the base for data gathering, the greater the incentive to subsidise the platforms which gather the data, creating strong pressures towards monopolisation.

2: the attaching of these advertising to core functional services - while televesion has been a dominant cultural medium in the 20th century, it still seems to be dwarfed in importance by the huge range of advertising-subsidized services, from a huge swathe of our communication in social media to important working tools create by Google, which between them allow Facebook and Google to dominate the entire digital advertising space.

One way of looking at this is as a response to a general difficulty of creating efficient markets where the creation of value is almost completely separated from the ability to distribute the product, as with journalism or music in which fixed costs dominate the production costs, and so with standard, marginal cost-based pricing, money will be lost and producers will quickly go out of business. However, sharing technology dictates that if the price to view is kept high, people will simply access it by other means. The solution has been to implement advertising which, if it imposes a cost on the individual, is a less salient one felt only at a later time.

The user pays nothing directly but the content creator gets paid. Where does this money come from? Is it paid simply out of the premium that individuals pay for brands that they have previously been exposed to? Does competition in advertising hurt producer surplus? Is there an overall utility gain from exposure to adverts?

I don't want to overstate the importance of this model - in many cases the ad-driven business model for services where companies begin by driving throughput and rely on ad revenue to sustain them has proved to be untenable in many cases. Nonetheless we see in Google and Facebook.two of the largest and most powerful corporations in the world whose power largely resides in their ability to collect data to generate a competitive advantage in supplying ads and so the phenomenon is clearly an important one.

Deriving a model.

How should we model this? The default paradigm in economic modelling tends to take consumers’ preferences as given but the existence of ads, which have the explicit aim of altering these preferences, we’re going to need to add more than this. So long as advertisements are at least minimally effective we will find that on average the post-impression preferences will shift the demand curve for the product outwards, and we have for a first decision whether we should pay more respect to the pre- or post impression preferences.

Firstly, we would expect that, at least in some cases, advertising represents extra information - this is a model that naturally seems more sympathetic to the advertising industry and applies best to companies who are expanding and presenting themselves to customers. The possibility of advertising is an important part of the free market model, as a channel for new competitors who believe they have a product better than the incumbents and want to let people know - and since the value of advertising should be highest to those companies that are in fact persuasive, especially with personalised advertising, we would expect there to be a gain in utility from the presence of advertising.

On the other hand, advertising is a major expense for a wide variety of very well established companies, with an extreme case study being Red Bull - essentially a energy drink that can be manufactured for a tenth of its retail price but which has, through extensive advertising, created a brand identity which allows it to charge a huge premium - the profits from which maintain its advertising push to sustain its position and so on.

Also, it is deeply ironic, and also a fundamental conflict of interest, that the development of Google and other search engines have radically increased our ability to evaluate for ourselves the range of possible product options and yet Google is also the world leader in facilitating paid advertising

One important dimension of this model that is rarely talked about but is central to our enjoyment of the modern web is that paying through viewing advertising approximates a form of first order price discrimination, something well known to be impractical in pretty much any other domain. This gives anyone with an internet connection the ability to access a wide range of free goods that have become accepted almost as birth right. The phrase ‘if you’re not paying for it, you are the product’ widely used to caution individuals using free products but it may very well be that this is a fair price to pay. Google, among other things, offers free mail, cloud storage and collaborative document editing, all working neatly together - in return for the ability to read everything that is stored on their servers. This tool suite, combined with the enormous wealth of educational materials available on YouTube, represents the capacity to educate oneself to a very high level at the cost of only the basic internet connection and the manipulation of one’s preferences - which is cost that scales with income and therefore is affordable to any user, in a way that ad-free pricing is not, and offering very cheap services the poor through explicit discounts . At the moment my entire expenditure is pretty much rent, food, beer and transport, none of which are seriously impacted by advertising. If at a later date I have far more spending money, there will be much more to gain by advertising products to me, but by definition I will be able to afford it.

The scale of an individual’s possible value to advertisers is, for those selling products, a function of their disposable income, which has a distribution even more extreme than simple income. For example, golf stars are often among the best paid athletes in the world despite having a global viewership that is dwarfed by football or basketball. They command these salaries because the average golf watcher is far wealthier than the average soccer fan, and so the endorsements surrounding golf are worth far more.

On the other hand it may be the case that lower income individuals are likely to be swayed by exposure to advertisements, because those with higher income are expected to be high information, with stronger norms aroudn and so are less swayed by advertising. The ability to sort through multiple trustworthy review sources may be the kind of data processing skill that is correlated with high incomes. The use of anti-advertising tools such as Ad-Blockers may also be something employed generally by high information individuals, and so if a high proportion of ad block users are higher income and higher information then this may not be as progressive as it first appears. I would have thought that young people would be much more likely to use them, but this report suggests that age is not a huge factor, with adoption rates varying from 22% for ages 25-34 to 15% for over 55s. The report suggests a strong connection with education but 26% of the survey pool obtained a degree compared to only 11% of the US population of that age. The survey thus is far from perfectly representative and they do not mention adjusting for demographics in their methodology so the absolute numbers I think must be taken with a pinch of salt.

When it comes to the holistic effects of advertising, one thing that would be extremely powerful to know and plausible to study would be: what is the effect of a general adblocker on overall spending, satisfaction with ones level of income,. If we could get a cross section of people, give them all adblocker for a year and see what effect this had on aggregate spending we would be somewhere closer to a real look at the macro effects. Can this be done?

Let us imagine that Google lost access to its current, ad/data driven model and could not justify maintaining these just for gateways to information and required payment for these services. On the one hand, this would make us lose out on important services, but on the other, lots of people would be pushed towards using open source versions of these. My own experience with these has been that they lacked the polish of Google and Microsoft’s proprietary software, perhaps reflecting the balance of developers to designers in the open source community, or perhaps coherent design is more difficult in the flatter structures of open source. Wikipedia, for example is a core part of these free tools that have come to be seem as birthright, but with volunteers and low bandwidth usage, it has managed to succeed without either ads or selling data. On the other hand, Wikipedia is able to work as a trustworthy source because it aggregates information published by journals and journalists across the world, most of which operate in an environment where writers are paid directly for their work. It seems likely these would take on a bigger global role in some areas if the only corporate alternatives were pay-to-access, but wherever significant uptime costs are necessary it seems these needs would be served by direct payment or not at all.

I wonder whether there exists a boost to utility to not having to ration one's consumption of a good, similar to going to a buffet. Most large online services used by consumers like Netflix and Spotify allow unlimited use for their users, but alternatives to an advertising model for smaller content publishers that involved microtransactions, paying for individual articles, for example, may decrease utility by forcing. On the other hand, it may be that when one pays for a specific piece of literature or journalism, the appreciation of it is greater. This is the difficulty of economic theorizing in a space whether we are forced to take conditions altering utility functions seriousl;y.

Possible policy responses:

There are likely to be some forms of advertising that are more beneficial than others. Those adverts which seek employees are likely to be more useful than those selling products, not just for those who are unemployed but switching jobs, and awareness of other opportunity in general tends to raise salaries for those in work. Differential taxation/subsidy for adverts with differing social consequences may be helpful in generating a pro-social advertising market.

Interestingly Ghose and Yang (2009) reckon and treat it as general knowledge that banner advertising is the most annoying form and use that to motivate their analysis of search engine, suggests that attitudes have changed, and as ads become pervasive there’s been a shift towards clearer, separated ads. Suggests public perception of advertising has been materially shifted by negative perception of Facebook style personalisation and integration of advertising (Need to find the research that I saw blogs using to justify their use of banner vs other ad types).

If we were to propose a theoretical model:

Interesting and valuable info on the power of t (also note, we like to think no income effects, but monthly payments for students likely to be serious, since their incomes are often calibrated to provide their basic needs and not much else. Also, game theoretically, individuals may pay a lot to stay at the co-operative/schelling point but 5that doesn’t necessarily mean that this point provides that much value over other points. A proposed model A cost added to the manufacturer for the advertising A term in the consumers preferences for how sensitive they are to advertising, a very important e^a*logx where a is the parameter of susceptibility to advertising or directed information on that product (is this valid? We could get a more stable situation by using like a scaled integral of the normal) x is the consumption of that good Y = sqrt(x) + a is the basic format i would expect, this is dangerous because we don't want to make the result simply depend on the relative order of the terms of the equations Possible Alternatives One proposed solution is the integration of microtransactions into . When looking at the value of these services, one thing to take into account is the feeling of unlimited use. From a simplistic model it may be hard to explain why buffets can exist because they are subject to adverse selection in the strongest possible way - why would anyone go to a buffet unless they expected to eat more than they could buy for the same amount outside of a buffet, leaving only those customers who would make the buffet unprofitable. One answer to this is that there is a psychological value in not having to ration one’s consumption according the cost of consumption. Another of course is that their food offerings are differentiated in more than just their price structure,. Slate star codex example, its pretty useful, imagine that, but done automatically for small fb groups. Neato! Also fair to think of the freedom of advertising, at the risk of assuming a capitalist mindset, pretty sure its reasonable to say