by Tim Hwang
Such a calibrated takedown may be more plausible than it initially appears. The erosion in trust necessary to trigger changes in the behavior of a marketplace may be smaller than you imagine. The stock market can enter a recession without every company in an economy experiencing a downturn. The subprime mortgage crisis didn’t necessarily depend on every mortgage defaulting or every mortgage being subprime. Similarly, the online advertising economy can head toward a crisis without causing the collapse of advertising as a whole. The decision of a few key players to cast their lot outside the modern programmatic ecosystem could very well be sufficient to bring the various looming problems we’ve discussed to the fore as it has in previous market bubbles.9
So how do we burst this bubble? Two pillars of faith give programmatic advertising an aura of invulnerability: measurability and effectiveness. The core proposition of programmatic advertising is that it gives advertisers an unprecedented depth of accurate data about consumers, which is able to produce uniquely effective outcomes for advertisers. For supporters, this makes programmatic advertising superior to earlier, established channels for running ads like television and print. For critics, this makes programmatic advertising a particularly powerful and pernicious force in society.
Reducing confidence in the measurability and effectiveness of programmatic advertising will chip away at the willingness of ad buyers to pour money into the ecosystem. This provides a brake that we can tap in order to slow the flow of cash into the marketplace. If it turns out that the data collected by advertisers are not in fact accurate, or that the data are not in the end all that useful for shaping perceptions and behavior, then programmatic advertising is not “better” in any real sense.
Independent research may be a particularly powerful tool for shaping industry views of online advertising. One of the reasons the programmatic advertising market is so overheated is that self-interested boosters dominate the ecosystem of information. Entities like the Interactive Advertising Bureau and the Association of National Advertisers are major outlets for research on the state of the marketplace, but they simultaneously serve as advocacy organizations on behalf of the industry. The space lacks a robust, independent institution to act as a counterweight, to objectively investigate industry claims and conduct ongoing experimentation to test the health of the marketplace.
The field of economics offers an interesting template. Founded in 1920, the National Bureau of Economic Research (NBER) is a “private, non-profit, non-partisan organization dedicated to conducting economic research and to disseminating research findings among academics, public policy makers, and business professionals.”10 The NBER’s first effort was a pioneering study measuring national income, a critical basic metric for measuring a “nation’s economic well-being [in] quantitative form,” which had previously been unknown.11 Over the subsequent decades, the data and research released by the NBER have played a major role in shaping thinking around the economy and its management.
Imagine an NBER for the advertising space; we’ll call it the National Bureau of Advertising Research (NBAR). The NBAR could conduct rigorous research on different aspects of the machinery of programmatic advertising, shedding light on previously opaque and murky elements and putting policy around the marketplace on firmer footing. This would allow elements of the bubble to be targeted selectively and focused on, tapping the brakes here and swinging a wrecking ball there as needed to reshape the size and structure of the marketplace.
Advertisers may prove to be allies and supporters of the NBAR, because they bear some of the biggest costs in the current attention marketplace. As the researcher Michael Kaplan notes, the pressure to compete for audience attention online and the declining value of programmatic ads means that the platforms “extract large price surpluses, effectively exploiting their customers [the advertisers].”12 The NBAR would provide those customers with a way of challenging the platforms and a basis on which to exit the market.
What would the NBAR investigate first? Nico Neumann’s claims—that the granular metrics and high accountability of programmatic advertising may not make any difference—might be a good place to start. Along those lines, the NBAR could investigate the intriguing anecdotal evidence that spending significantly less on online advertising actually improves outcomes for the advertiser. In 2017, Procter & Gamble slashed around $200 million of its advertising spending in the digital space, citing concerns about bot fraud and brand safety. This money was reinvested in more traditional advertising channels like television and radio. The outcome? P&G reduced its overall spending and still increased the reach of its messaging by 10 percent.13 More concrete exploration on this front might pressure the market to price the value of online advertising accordingly.
Similar patterns have appeared among sellers of online advertising as well. In response to the new General Data Protection Regulation privacy rules in Europe, The New York Times shut off all open-exchange programmatic buying on its European pages in 2018, preventing ad buyers from using the rich data available about specific consumers. In theory, this should have produced a decline in the demand for the Times’ ad inventory; ad buyers would flow to places with more user data to sell. But the paper’s digital advertising revenue was not affected at all. The Times “briefly tested reintroducing open-exchange programmatic ad buying … but didn’t pursue it.”14 This phenomenon has been analyzed by Dr. Alessandro Acquisti at Carnegie Mellon University, who concludes from his work that it is unclear that data-driven targeted advertising actually produces significant returns. Acquisti’s team found that although targeted advertising increased revenues for publishers, it did so at the barely noticeable rate of about “$0.000008 per advertisement.”15
Well-grounded, hard-hitting research is powerful: it will help to persuade advertisers and publishers to leave the programmatic marketplace, and may even shape the behavior of consumers at large—say, through accelerating the adoption of ad-blocking technologies or encouraging people to switch to alternative platforms not based on advertising. All this would help produce a managed crisis in subprime attention.
But we cannot be confident that the NBAR alone would bring down the programmatic marketplace, given how profitable a willful ignorance about programmatic advertising can be. The effort to help ground public and policy-maker understanding of the internet’s attention marketplaces can and must be complemented by more aggressive action from activists and whistleblowers. This could take the form of a more aggressive style of research, demonstrating the vulnerabilities of the marketplace with direct action: distributing scads of faulty consumer data in order to evaluate the lack of quality control among those buying and selling these assets, or running coordinated algorithmic bidding scripts to identify how the programmatic auction system can be manipulated. Leaks can reveal the extent to which platforms, agencies, and other actors in the ecosystem are behaving badly, and simultaneously keep those actors on their toes. Anonymous online submission systems for whistleblowers to securely leak documents around wrongdoing in the digital advertising industry could be a good way to encourage this activity. These more provocative actions could help to adversarially target and demonstrate the vulnerabilities of more intransigent actors that would otherwise resist their profits shrinking.
As in the financial markets, governments and regulations will have an important role to play. The success of even a relatively tame version of the NBAR would depend on the ability of that organization to accurately assess the state of the marketplace and to conduct experiments that truly scrutinize the claims of the industry. Vested interests within these marketplaces are not likely to grant such access willingly. In the past, access to the inner workings of the programmatic advertising infrastructure has been only grudgingly given by companies after the industry has come under public criticism and business pressure. More consistent disclosure will be critical to the project of a graduated dismantling of the subprime attention bubble.
Disclosure and Stability
The toxi
c blend of eroding value, market opacity, and bad incentives is producing an unsustainable market due for a painful correction. Self-regulation is failing to resolve these problems in a meaningful way and bad incentives will hinder serious efforts to clean these marketplaces up. Moreover, the intertwining of the online advertising marketplace with the fate of the broader economy means that the public has a stake in the stability and operation of the business of buying and selling attention. These facts suggest an evolving role for the government in shaping the rules of play around programmatic advertising. But what form should this take?
The similarities between the financial markets and the attention markets motivate much of this book. The history of the financial markets provides a number of useful parallels to understanding the evolution of online advertising and suggests how ostensibly limitless engines of profit can be in reality quite unstable. But the long history of thinking about financial markets and their management also gives us another important thing: a grounded sense of how regulation can help to mitigate excesses in precisely the types of markets that online advertising now resembles.
While the subprime mortgage crisis has been our primary touchstone for the last few chapters, 1929 might serve as a more apt analogy than 2008 in thinking through the question of what regulation should look like. The year 1929 was the year of the Great Crash, a catastrophic collapse in the value of stocks and other securities traded on the New York Stock Exchange. Among economists, the Great Crash is historically significant because it marks the beginning of the Great Depression, which lasted for the next twelve years and represented a colossal 15 percent decline in global GDP.16
These events prompted the U.S. Senate Committee on Banking and Currency to investigate the crash of the New York Stock Exchange. Later known as the Pecora Commission after its chief counsel, Ferdinand Pecora, the investigation revealed a broad array of fraudulent activities taking place within the financial sector in the years leading up to the crisis. The Pecora Commission’s findings provoked a raft of new regulations during the opening days of the Roosevelt administration that fundamentally changed the operation of the financial markets in the United States.
The less-than-savory activities uncovered by the Pecora Commission will sound oddly familiar to observers of the programmatic advertising markets. Pecora found that investment banks willingly sold gigantic lots of securities on behalf of businesses and governments known to be on shaky financial footing.17 He also investigated the use of so-called preferred lists, which provided gifts of steeply discounted securities to a circle of influential individuals who could lend aid to the banks.18 Moreover, Pecora highlighted the role that flimsy state-level regulations played in enabling a murky environment that made it impossible for buyers of securities to assess precisely what they were purchasing.19 It was, in short, the same set of elements present in programmatic advertising: a potent blend of market opacity, toxic assets, and conflicted players that inflated an enormous bubble in securities, which eventually burst to catastrophic effect.
The financial giants investigated by Pecora were the technology giants of their era, in terms of both their economic importance and their cultural prestige. Investment banks investigated by the commission, like J. P. Morgan & Co., were businesses with small numbers of employees that generated huge wealth. Time magazine described the banks as “the greatest and most legendary private business[es] of modern times.”20 Banks like J. P. Morgan’s—“the most famous and powerful in the whole world”—were condemned by the press as “fail[ing] under a test of [their] pride and prestige.”21 To the anger of the public, Pecora also exposed the tendency of these businesses to pay little or no tax, a controversy echoed in modern-day investigations into the global online advertising platforms.22
Importantly, the Pecora Commission shone light on the critical role that a lack of transparency played in bringing about the Great Crash. Buyers of securities simply could not get a full picture of what they were purchasing or of the conflicting interests that the sellers might have had in the transaction. Blue-sky laws, which mandated these disclosures, operated at the state level and were easily circumvented.23 Rules mandated by stock exchanges were voluntary and similarly easy to evade.24
The commission inspired the passage of the Securities Act of 1933, which tackled the transparency issue head-on. Critically, the act established a “disclosure philosophy” that is even today “generally regarded as the appropriate or inevitable method of regulating corporate finance.”25 In brief, the act set up a process by which certain significant categories of securities were to be “registered” with the Federal Trade Commission—later the Securities and Exchange Commission—prior to being sold. The registration process requires the issuer of a security to disclose to the public a wide range of information about the company. This includes facts about its controlling board, financials, and other key information. Individuals who sign registration statements on behalf of the company issuing securities are held legally liable for any inaccuracies in the registration statement. The assurance that these documents were signed under this threat, in turn, is designed to support investor confidence in the viability of the assets being traded in the marketplace. This basic registration framework still governs the issuance of stocks and other securities within the United States today.
The disclosure philosophy is modest in some respects. It is designed to ensure that accurate information is available, but it does not ensure that a given security is a good investment. As Franklin Delano Roosevelt wrote in a message accompanying the introduction of the 1933 act, “The Federal Government cannot and should not take any action which might be construed as approving or guaranteeing that newly issued securities are sound … There is, however, an obligation upon us to insist that every issue of new securities to be sold in interstate commerce shall be accompanied by full publicity and information … It puts the burden of telling the whole truth on the seller.”26
In other words, the health of a marketplace does not depend on every asset being worthwhile. The Great Crash teaches that markets do not panic simply because business is bad. Panics are instead a product of the lack of trust that buyers and sellers put on information circulating in the marketplace. Opacity allows for bad behavior on the part of conflicted players and creates uncertainty about what information is reliable. These factors permit expectations within a marketplace to float far from reality before being suddenly brought back to earth. Simply put, transparency enables market stability.
The 1933 act remains a bedrock of securities law in the United States, despite the massive transformations within the economy over the decades since it was passed. While it has by no means been immune to criticism or prevented every financial crisis, there is an overall consensus that the disclosure arrangement it sets up is an important element of a stable securities market. One scholar reviewing the empirical literature on the impact of mandatory disclosure concluded that “the effects of mandatory disclosure on stock returns, volatility and financial development are consistent with mandatory disclosure often having socially beneficial effects.”27
Something like the disclosure philosophy may be a critical piece of the puzzle in grounding the programmatic advertising markets and deflating the bubble. The 1933 act confronted a situation in the financial markets that matches the murky, highly conflicted circumstances of the online advertising markets. The boosters of online advertising have long pushed the notion that the internet is transparent and trackable in a way that earlier generations of advertising were not. But these claims fail to tell the whole story. Online advertising has introduced new blind spots and failed to address some long-standing ones. These have allowed a range of frailties to creep into the marketplace that threaten its long-term stability.
Demonstrably, the programmatic advertising market lacks sufficient incentives for candor. The programmatic advertising industry has remained relatively opaque about the state of the overall marketplace, despite demands from ad buyers and some embarrassing i
nvestigations. Ironically, this prevailing refusal to reveal critical data about the state of programmatic advertising to the public contributes to the risk of a catastrophic implosion. Without consistent rules governing the release of these data, buyers and sellers are left reacting to a mess of partial, easily misinterpreted information that is disclosed only when doubts already loom around the veracity of claims made about the marketplace. The likely result is overreaction and potentially cascading losses of confidence.
As in 1933, ensuring that a market reflects reality may require mandating a higher level of disclosure than the industry left to its own devices would provide. Mandated disclosure also empowers research organizations to examine the state of the marketplace and conduct experiments, giving policy makers and the public a better grasp of the economic health of the internet.
So how would the disclosure philosophy of the 1933 act apply to the world of programmatic advertising? The basics would be the same: prior to offering ad inventory for sale in the programmatic marketplace, sellers of attention would be legally required to provide to the public a standardized statement of relevant information. This statement of information would shine a spotlight on precisely the areas we have covered in the course of this book. It might include detailed metrics around brand safety, performance, proof of business relationships, disclosures of conflicts, and so on.
The disclosure mandate will need teeth: these representations can and should put the company selling ad inventory on the hook if they turn out to be inaccurate later. Legal liability is perhaps the heaviest (and slowest) deterrent, but a range of potential punishments are available to those who falsify or otherwise spread misleading information in these disclosures. One might be excluded from selling advertising inventory across certain marketplaces or be clearly marked as a seller with a bad record. These disincentives would deter attempts to obfuscate the value of advertising being bought and sold.