by Tim Hwang
Mortgage originators were incentivized to make riskier and riskier subprime loans as the demand for these assets continued to grow. As a percentage of all mortgages, subprime mortgages rose from 8 percent of the market in 2003 to 20 percent in 2005.3 Those mortgages would go bust en masse, eventually triggering the broader financial crisis.
The key lesson of the savings glut hypothesis is that large flows of money seeking safe harbors can produce the conditions for a bubble to form. After decades of financial instability, governments in the developing world sought safe places to put their money. As the price of the safest option—U.S. government debt—rose, mortgages appeared to many financial institutions to offer an ideal combination of affordability and reliability. Those structural flows drove the market up even as warning signs began to emerge that the mortgages being issued were nowhere near as stable as was originally thought.
The savings glut hypothesis is controversial. Some argue that it understates the failure of financial regulators to address the warning signs that presaged a broader crisis. The journalist Martin Wolf says the hypothesis “may have had the right analysis, but the Fed’s was the wrong response, made worse by the failure to regulate the financial system.”4 Like many complex social and economic phenomena, no single cause brought about the 2008 subprime crisis. However, as we think about online advertising markets and their fragility, the savings glut hypothesis provides a provocative argument for explaining why ad markets may continue to grow despite their known weakness.
We have been diving deep into the programmatic advertising marketplace for the last few chapters, but it is worth thinking about the broader landscape that an ad buyer faces. The ad buyer has to distribute a limited budget as effectively as possible across the various channels available to her. This might involve buying ads through the methods we’ve been focused on: bidding to distribute ads through platforms like Facebook and Google, or buying an ad through an exchange. But it might also include older, established media: buying an ad in a newspaper, or arranging for a television spot, or renting billboards in major cities around the country.
All of these channels provide a certain level of perceived return to the ad buyer. This might be measured in extremely concrete ways—purchases of a product—or in softer ways that reflect a more brand-advertising-based approach. An advertiser might find it valuable to expose a product to an audience that is unfamiliar with it, regardless of the immediate purchases they choose to make. The ad buyer is an investor of a sort, choosing how to allocate limited resources across many different types of promotions that offer different kinds of returns.
In this context, online platforms like Facebook and Google offer new, shiny assets in the broader marketplace for grabbing attention. Google and Facebook do not simply compete with one another for ad dollars; they also compete against the many other kinds of businesses that offer slices of attention for a fee. Digital advertising is just one kind of investment in the marketplace for attention. Ad buyers can choose to spend their limited dollars on digital instead of, say, television, hoping for a greater proportional return.
And for the most part, they have done just that. The meteoric growth of online advertising has come at the expense of traditional attention channels.5 Classified ads, the old advertising workhorse that funded many a local newspaper, were outcompeted by platforms like Craigslist and Google.6 Globally, digital ads are a $273 billion business, but this is just a subset of the bigger industry of advertising. Overall ad spending across all media in 2018 was $629 billion. Digital is, however, poised to take over more and more of that overall amount, and in 2019 projections showed that fully half of all ad spending would go to digital by 2020.7
Digital companies see a future in extracting the advertising dollars still currently invested in older, established channels for distributing messages. This explains in part Facebook’s pivot to video and Google’s continued investment in YouTube. The successful capture of even part of the money that is spent on television ads would bring in billions for these platforms.8
But this is a two-sided game. Facebook and Google do not just compete with television networks and billboard companies for the dollars that advertisers are looking to spend. They are competing with these other advertising channels for public attention, too. The internet captures some of the finite attention that other media channels are vying for as well. People can watch YouTube instead of broadcast television or choose to read blogs over newspapers.
This makes digital advertising a double threat. On the one hand, digital advertisers compete against other media channels for advertising dollars. Coca-Cola, for instance, might choose to put advertising dollars into promoting its latest soft drink on Facebook rather than airing a commercial on CNN. On the other hand, digital advertisers also compete for the viewers that make legacy media channels attractive to advertisers in the first place. Consumers increasingly spend time on Facebook that they might otherwise spend watching CNN. Facebook thus appears more valuable to the advertiser. As advertisers give less money to media like newspapers, they also cripple the ability of newspapers to offer the content that subscribers demand. This makes it harder for legacy media companies to compete against their digital rivals.
This structural shift has far-reaching implications. Consider for a moment what effect this death spiral of legacy media has on ad buyers. It is not just that established media options are less attractive but that over the medium to long term they are increasingly unavailable as options at all.
The movement of consumers to digital platforms makes it riskier and less attractive for ad buyers to buy into anything but digital advertising. The shift of advertising dollars onto online platforms accelerates the deterioration and disappearance of competing outlets for distributing ads. The disappearance of those competing outlets, in turn, accelerates the flow of money onto online platforms. Increasingly, all roads lead to the programmatic advertising economy.
This structural shift looks a lot like the savings glut. Like developing world governments in the 1990s and early 2000s, advertisers confront a media ecosystem where traditional media outlets are looking shaky or are disappearing outright. This produces a huge influx of cash into the online advertising marketplace that has nowhere else to run. Some of that cash will end up in relatively safe places: the U.S. Treasuries of the online advertising world. Some of it will not, particularly as the limited number of safe, high-value opportunities rise in price and disappear into private marketplaces.
Under these circumstances, it would be difficult for the online advertising ecosystem to correct itself even if it had the will to do so. Ad buyers are unlikely to cut their budgets significantly even as the number of options for placing those budgets declines. That provides sufficient fuel for a bubble, despite the persistent problems of opacity and subprime attention.
But feeding that immense level of demand can also take on some perverse characteristics. Loan originators creating mortgages to meet global demand during the 2000s eventually found themselves under intense pressure to look the other way as the quality of those mortgages became shoddier and shoddier. Similarly, the marketing agencies and ad technology companies profiting most handsomely from these developments have few incentives to address the market’s deep structural flaws. In some cases, they may even exacerbate these issues.
Perverse Incentives in Financial Markets
Online advertising inventory is increasingly dubious, whether it is ignored, blocked, or entirely fake. But a decline in value is not, in itself, a problem for a well-functioning marketplace. Buyers can simply refuse to pay a premium for online ads, driving the price of ad inventory downward. While the total value of advertising sold in the marketplace would decline, the market wouldn’t necessarily implode dramatically. Prices can be a powerful means of averting market crisis. The market simply incorporates the risks of things like ad blocking and the generally lower level of attention paid to display inventory, and life goes on.
But if price fails to r
eflect reality, the price of the ads being bought and sold will remain stable as the real value of the underlying asset—attention—erodes. That raises the risk that these prices might suddenly “snap” to their real value amid a loss of confidence in what is being bought and sold.
How might this happen? We’ve already talked about one way. Market opacity can cripple the ability of price to accurately reflect value. Simply stated, if it’s impossible to see the underlying assets in a marketplace, it is impossible for the market to price those assets accurately.
Another major contributing factor are the players in the marketplace. A potent brew of greed and insulation from consequences can encourage players to stoke the market and overvalue what is being traded. A critical mass of these bad actors will inflate the bubble and prevent the market from adjusting effectively over time. The record of the 2008 mortgage crisis is replete with examples of banks and rating agencies that took reckless bets and hid the true vulnerabilities of underlying assets.
On the bank side, mortgage-backed securities were traded on an “originate and distribute” basis, which meant that the institutions responsible for creating these assets did not hold them on their balance sheets. Instead, these assets were sold off into the marketplace, where they were traded among a large number of buyers and sellers. Because the originating banks were compensated by volume, bankers had perverse incentives to originate and package as many assets as possible to meet demand. In response to this pressure, banks began issuing mortgages to everyone who applied. Infamously, borrowers were able to obtain high levels of financing for purchasing homes with little money down and with no information required about their job or finances—so-called NINJA (no income, job, or assets) mortgages.9
As the market for mortgage-backed securities grew, the assets were increasingly filled with highly toxic, low-quality mortgages that were unlikely to ever be paid back. This problem was compounded by the Big Three of global credit ratings agencies—Standard & Poor’s, Moody’s, and Fitch Ratings—which were responsible for assigning grades to the packaged securities. The investment banks issuing these assets paid the ratings agencies, perversely incentivizing the agencies to give the banks’ toxic mortgages high AAA ratings.10 The high ratings fueled the market for these assets and obscured the real, underlying fragility of mortgage-backed securities. AAA ratings also enabled massive money market and pension funds, restricted from holding lower-rated assets, to purchase mortgage-backed securities in great numbers.
The incentive to overvalue mortgage-backed securities laid the groundwork for a larger crisis, because it was impossible to hide the underlying weakness of these assets indefinitely. Many of the mortgages packaged into these securities came with a “balloon rate” that would rapidly increase the interest due on the mortgages a few years after issuance. As these triggered in increasingly large numbers between 2006 and 2009, $738 billion in mortgages suffered “payment shock” and default.11 The increase in mortgage failures cast doubt on the purported value of these assets, eventually triggering a widespread panic.
Similar conflicts pervade the modern marketing ecosystem, spurring the expansion of the programmatic market bubble even as the problems continue to grow. Marketing agencies and advertising technology companies play the role of the pre-crisis ratings agencies and loan originators. The business practices of these entities juice the market in ways that assist the growth of a bubble.
Perverse Incentives: Agencies and Ad Tech
Market bubbles are seldom a surprise to everyone. Industry insiders can buoy the growth of a marketplace even as the underlying fundamentals grow worse and worse. In the 2008 crisis, financial institutions and ratings agencies profited from the ever-expanding demand for mortgage-backed securities even while they were aware that the market was more fragile than it looked from the outside.12
In the online advertising marketplace, two groups have systemic incentives to oversell the value and price of advertising inventory regardless of issues like fraud, declining attention, and ad blocking. There are the marketing agencies, seeking profits in an increasingly unfriendly business environment. And there are the marketplaces themselves, which will cease to be profitable if people stop buying and selling online attention. These actors work to prop up the price of digital advertising inventory in spite of its eroding value.
The Agencies
Marketing agencies have long played an influential role as the middlemen between buyers and sellers of advertising inventory. This privileged role has come under competitive pressure in recent years as the rise of programmatic marketplaces has facilitated a more direct ecosystem of transactions between advertisers and publishers.13 Functions previously tasked to agencies are now performed by other actors. For instance, creative development of advertising content has increasingly been brought in-house. Large advertisers have built their own internal agencies, and major publishers like Facebook and Google provide creative services directly to large advertisers.14
Financial pressure from clients is also rising. Clients are closely scrutinizing their return on investment and demanding greater transparency. From 2015 to 2018, major brands—including Shell, HSBC, and Mars—undertook top-to-bottom reviews of their marketing budget allocation.15 This wave of budget reviews led to the cancellation of many long-standing contracts between clients and marketing agencies.16 Marketing agencies face ever-shrinking profit margins in light of this loss of market position. This has left agencies scrambling to find other ways of making money in this new environment.
One controversial means of doing so is manipulating the price of advertising inventory through an industry practice known as arbitrage. Marketing agencies sign deals for discounted prices for ad inventory from publishers. These agencies then turn around and resell this inventory at a higher price to their clients, pocketing the difference. Similar practices exist in the acquisition and selling of consumer data by agencies to advertiser clients.17
Arbitrage is a major concern in the advertising industry because it potentially places the agency in conflict with the advertiser it supposedly represents. For one, the discounted nature of the ad inventory being acquired might introduce bias, encouraging agencies to recommend using certain channels to distribute advertising that they otherwise would not. Second, arbitrage adds additional unnecessary cost to advertising budgets because the agency charges a markup with no justification. These costs would be avoidable if ads were easily acquired in the programmatic marketplace.
According to one industry veteran, these practices are “happening virtually everywhere in the U.S. media landscape.”18 A major independent study commissioned by the Association of National Advertisers (ANA) in 2016 concluded that “non-transparent business practices were found to be pervasive.” It went on to observe that “these practices appeared to be part of the regular course of business” across a cross section of major agencies and agency holding companies interviewed for the report.19 One former executive stated that the entire system of agencies and clients was “one massive arbitrage system.”20
Agencies remain a significant part of the digital advertising ecosystem and play a major role in coordinating advertising campaigns and spending. The prevalence of arbitrage means that these agencies have a strong incentive to aggressively inflate the value of the inventory they are selling. This inflates the market bubble even as the value of this ad inventory declines. Like the ratings agencies in the march to the subprime mortgage crisis, marketing agencies have conflicted incentives to give figurative AAA ratings to low-value ad inventory, because they profit directly from doing so.
This may explain why, despite growing concern from advertisers and its negative impact on major agencies, the arbitrage conflict has not been addressed in a serious way. In 2018, two years after the ANA report, the industry publication Digiday opined that “advertisers and agencies talk about transparency a lot but often look as though they’d rather blame one another for the lack of clarity than come up with a way to get it.” It quoted a
representative of the ANA who noted that “trust between advertisers and agencies is lower than it’s ever been because agencies keep denying that there are transparency issues [around arbitrage].”21
Ad Technology
The opacity introduced by the speed and scale of programmatic exchanges doesn’t just make it more difficult to know where ads will end up online. It also makes it hard to determine why a particular unit of ad inventory commands the price that it does. Ad technology platforms, we will see, inflate prices in very much the same way that marketing agencies do.
As we discussed earlier, the programmatic marketplace relies on the interaction between two types of platforms. On the one hand, there are the demand-side platforms (DSPs), which help advertisers purchase ad inventory. On the other, there are the supply-side platforms (SSPs), which contract with publishers and obtain the privilege of offering their ad inventory to buyers.
These platforms are not neutral players in the marketplace. DSPs and SSPs have been known to work together to inflate ad prices in order to secure profit for themselves. Controversially, these practices are not typically disclosed to the advertisers and publishers whose interests the platforms ostensibly serve.
DSPs can sign bulk purchase deals to acquire inventory from their supply-side counterparts at a discounted price. This discounted inventory is then sold by the DSP at a huge markup with no disclosure of the increased cost. These margins, which are occasionally disclosed in the public Securities and Exchange Commission (SEC) filings of some DSPs, can range from 44 percent to 66 percent.22
SSPs, for their part, have in a number of cases misrepresented the relationships they have with various publishers and the types of inventory they are authorized to sell on their behalf. When a buyer goes to purchase this inventory, the platform simply resells inventory purchased from other SSPs.23 In these cases, the SSP is an unnecessary middleman, artificially inflating the price paid by the ultimate buyer.
