If this conclusion sounds overly bearish, it is probably in reaction to the unwarranted hype that these companies have generated. There is no doubt in my mind that the GAFAMs are genuinely impressive companies, but so were GM, GE, IBM, and AT&T before them. They are not special and should be treated with the respect and circumspection other companies receive.
If there is one lesson that economic history teaches us, it is that great companies need to be challenged (and that one definitely applies to all the GAFAMs, Amazon included). We do not know for sure why productivity is slowing. Perhaps ideas are becoming harder to find, as Bloom et al. (2017) argue. But this book proposes that declining competition and rising barriers to entry have allowed incumbents to rest on their laurels. We need to bring in more competition. The problem, in the case of internet firms, is to find the right tools to do so.
* * *
a The FTC approved a merger between Exxon and Mobil in late 1999.
b There might be a broader lesson here, but the analysis is preliminary. OECD researchers Dan Andrews, Chiara Criscuolo, and Peter Gal (2015) have studied frontier firms using harmonized cross-country data. They define global frontier firms as the top 5 percent of firms in terms of labor productivity or multifactor productivity levels within each two-digit industry in each year since the early 2000s. Global frontier firms are more productive by definition. They are also more capital intensive, larger, more profitable, and have more patents. They are also more likely to be part of a multinational group. They argue that the productivity slowdown of the past twenty years is not due to slower growth at the frontier but rather to an increasing productivity divergence between the global frontier and the rest. Between 2001 and 2013 average labor productivity at the global frontier grew at an average annual rate of 2.8 percent in the manufacturing sector and 3.6 percent in the market services sector, while the corresponding growth rate of all other firms was around 0.5 percent in both sectors. What they call frontier firms, however, are not GAFAMs at all. In their sample, the average revenue of “frontier” firms is around $40 million in manufacturing and $5 million in services.
CHAPTER 14
To Regulate or Not to Regulate, That Is the Question
The FTC should tread carefully when reviewing Google, Facebook, Twitter or any other tech company, given the dynamism of our tech industry and the potential for making things worse through regulation.
JARED POLIS
IN 2012, at least thirteen members of the US Congress sent letters to the Federal Trade Commission concerning its investigation of Google. Some of the letters reflected the belief that, like banks too big to fail, one of the country’s five biggest tech firms was too important to be investigated. US Representative Jared Polis, a Democrat from Colorado, voiced his concern “that application of anti-trust against Google would be a woefully misguided step that would threaten the very integrity of our anti-trust system, and could ultimately lead to Congressional action resulting in a reduction in the ability of the FTC to enforce critical anti-trust protections.”
In the previous chapter we analyzed the business models and reach of these firms: Google, Amazon, Facebook, Apple, and Microsoft—the GAFAMs. I have argued that the stars of the digital economy are not as special as people think. Or, to be more precise, they are not special for the reasons that most people think. They are not the pillars of the US economy. Their profit margins and market values are in line with historical norms. What is new is that they have a smaller footprint in the real economy than previous vintages of stars. If we exclude Amazon, the defining feature of the new stars is how few people they employ and how little they buy from other firms.
In this chapter we look at the topics that make the GAFAMs controversial—lobbying, tax evasion, privacy, and antitrust—and the extent to which these issues do or do not create barriers to free entry. We will also analyze how big data is used and consider the not-so-obvious role of price discrimination.
The GAFAMs Go to Washington
Federal authorities, and Congress in particular, were originally somewhat gun-shy about trying to control the burgeoning tech industry. Lawmakers resisted calls to force online retailers to collect sales taxes and generally stayed out of the way while online startups like Amazon, Google, and Facebook took the first steps toward growing into the behemoths they are today.
Faced with a relatively low risk of serious interference from Washington, the largest tech firms did not follow the example of other major industries by building up a large lobbying presence in the capital. Amazon, founded in 1994, didn’t begin spending money to lobby policy makers until 2005, and Google, founded in 1998, stayed out of the game until 2006. Founded in 2003, Facebook didn’t begin investing in a heavy federal lobbying presence until 2013. Apple, in business since 1976, had a de minimus approach to lobbying until about 2014.
In the past decade, though, the GAFAMs have gone to Washington. They have rapidly increased their lobbying expenditures, investing about $50 million in 2017. They lobby about immigration, net neutrality, rules governing advertising, and company-specific issues.
Figure 14.1 shows that the GAFAMs’ lobbying efforts are recent. The exception that proves the rule is Microsoft. In 1998, the DoJ brought antitrust charges against the software giant over its bundling of Internet Explorer with the Windows operating system, which, it was alleged, was a transparent effort to crush innovative start-up Netscape, which had created the Navigator web browser. The lawsuit took two years to settle and nearly ended with Microsoft being split up by a federal judge. As Figure 14.1 suggests, Microsoft’s experience with the DoJ in the late 1990s seemed to convince the company of the importance of having a presence in Washington, DC, and its lobbying expenditures have been steady since then.
FIGURE 14.1 Lobbying expenditures. Source: Center for Responsive Politics
Why did the other GAFAMs suddenly feel the need to hire lobbyists? As we saw in Chapter 9, companies typically increase their lobbying efforts precisely because they feel threatened, or at least potentially threatened. As the GAFAMs’ dominant positions became more obvious, and amid a string of scandals related to their treatment of users’ data, they began to attract more regulatory scrutiny.
Amazon’s lobbying increased after its acquisition of grocery chain Whole Foods. Facebook has been embroiled in a string of data privacy scandals, one of them involving the firm Cambridge Analytica. Waymo, Google’s self-driving car unit, faces potential liability issues and other concerns. Google, Twitter, and Facebook are also involved in the targeting of their users by Russian agents during the 2016 campaign.
Generally, companies exert influence in Washington for one of four main reasons. The first two reasons are related to benefits they expect to receive thanks to their lobbying efforts. They want either to protect a privilege they already have or to convince policy makers to bestow one that they don’t have yet. The other two reasons are related to costs they hope to avoid. Companies lobby to convince policy makers to lift an existing burden or to prevent them from imposing a new one.
We’re going to start with an issue that is relevant to all corporations: taxes. Then we are going to focus on one particular benefit that tech companies currently enjoy: massive concentration and network effects. We’ll also consider one burden they are anxious to avoid: new privacy protections for their users.
Do the GAFAMs Pay Their Taxes?
No, the GAFAMs don’t really pay their fair share of taxes, but to be honest, neither do the other global firms. All top companies have been paying fewer taxes over time. Figure 14.2 shows the taxes paid by large firms relative to their operating income. Effective corporate tax rates have decreased over time. They were around 50 percent of operating income until 1980 and then decreased to less than 20 percent. The tax rate of the GAFAMs follows exactly the same path as the tax rate of the other leading firms. In case you are wondering, the massive decline in 2016 seen in Figure 14.2 is due to an anomaly in ExxonMobil reporting related to an oil price dr
op.
The issue of corporate income taxes is a rather complicated one. According to standard economic theory, it is a bad idea to tax corporate profits. It is usually more efficient to tax distributions to investors—interests, dividends, and capital gains—because corporate taxes are more likely to reduce investment.
That’s in theory. What about in practice? The evidence is supportive of the standard argument, but not overwhelming. In general, researchers have found a negative impact of corporate taxes on investment, but the magnitude of the effects varies a lot across studies. Alan J. Auerbach (2002) provides a survey of these studies, and Simeon Djankov and co-authors (2010) look at more recent evidence. Nonetheless, there is some consensus among economists that corporate taxes should not be too high, and, more important, that they should be broad-based and free of loopholes.
FIGURE 14.2 Corporate income tax rates. Total reported taxes over operating income.
In recent years, the main issue has been corporate tax evasion, which is legal for the most part but costly and inefficient nonetheless. According to research by Berkeley economist Gabriel Zucman, the US loses around $70 billion in tax revenue each year because corporations shift their profits to tax havens.a That is almost one-fifth of all corporate tax revenue. As Zucman explains, almost two-thirds of “all the profits made outside of the United States by American multinationals are now reported in six low- or zero-tax countries: the Netherlands, Bermuda, Luxembourg, Ireland, Singapore and Switzerland.”
The GAFAMs do not seem to pay lower tax rates than other top companies. Large pharmaceutical, finance, and manufacturing companies engage in about as much profit shifting and tax evasion as the GAFAMs.
There is one issue that is more prevalent with the GAFAMs: it is often harder to locate the profits. This is what makes the GAFAMs’ tax issue politically explosive in Europe. One proposal is to compute taxes based on revenues instead of reported profits. If a company has high turnover in a particular country, it would pay more of its taxes to that country, even if it does not report high profits there.
In 2018, the US passed the Tax Cuts and Jobs Act (TCJA), which changed how international profits are taxed. It reduced the statutory corporate tax rate from 35 percent to 21 percent. This is not such a large change once you realize that tax breaks and loopholes had already reduced the effective rate much below 35 percent before 2018. The Institute on Taxation and Economic Policy showed that Fortune 500 companies paid an average federal tax rate of 21.2 percent between 2008 and 2015. But closing loopholes and lowering the statutory rate is probably a good idea.
The TCJA contains many provisions, and it is too early to tell how well it will work (see Chalk et al., 2018, for an early appraisal). It has, however, affected the way firms report and provision their taxes. Before 2018, firms had a choice: they could either provision their taxes on offshore earnings and include them in their headline taxes, or they could consider these earnings as perpetually reinvested outside of the US and not provision anything. Apple chose to provision taxes on a large fraction of its offshore earnings, which is why its headline tax rate appears relatively high. At the same time Apple lobbied hard to make sure it would never actually need to pay these taxes. The TCJA vindicated these efforts with a repatriation provision that taxes foreign earnings at a reduced rate of 15.5 percent, which is less than half of the rate used to provision the taxes. As a result, a large share of the taxes provisioned by Apple before 2018 will never be paid. All large companies engage in profit shifting, and most make use of tax havens to avoid paying taxes. But booking taxes that one does not expect to pay seems particularly disingenuous.
To wrap up our discussion of taxes, it is important to emphasize that corporate tax evasion represents a clear failure of public policy. Corporate tax evasion is mostly legal and could be reduced or even eliminated with a minimum of political willpower. Tax dodging by wealthy households, on the other hand, is mostly illegal and a more difficult problem to solve.
Is Concentration Required in the Digital Economy?
Information technology (IT) markets are highly concentrated. To understand the benefits of concentration to the GAFAMs, one must understand the principles of network economics. If we set aside political economy and lobbying, two economic forces can explain why and how a market becomes concentrated. The first explanation is that there are economies of scale: IT businesses have large fixed costs and then small marginal costs. This is an old idea, and not specific to IT. It also applies to pharmaceutical companies or aircraft manufacturers. Economies of scale might be stronger in IT because the marginal cost of information diffusion is often small or even zero, but that is an empirical question.
The second explanation—network effects, or externalities in the jargon of economics—is more specific to IT. The larger the network, the more members of the network have a chance to interact with each other. Reciprocally, people who are not on the same network cannot easily interact. Box 14.1 explains the basics of network economics.
Network externalities are a form of synergy. A positive synergy occurs when the sum is greater than the parts. When two firms merge, they might be able to reduce their costs by combining their IT systems, their human resource management, and other functions. They might create better products by combining their technologies.
Network externalities themselves come in two flavors: direct and indirect. The Facebook network is an example of a direct externality. We value Facebook directly because our friends are also on Facebook. In general, direct externalities arise when we derive utility from being in direct contact with another person on the network.
Indirect externalities arise when the presence of other users leads the network to offer services that we also enjoy. We do not care directly about being in touch with the other people on the network, but we share similar interests. A good example is the ecosystem of apps developed for an operating system, whether they be on a phone, a tablet, or a computer. In that example, we are not planning to interact directly with the other users. But when the number of users grows, developers have an incentive to build more apps, and we all benefit from them. At the end of the day, even though we do not value direct interactions with other users, we still value their presence on the network. A network with few users would have too few apps to be attractive to us.
Box 14.1. Network Economics
Two concepts play a key role in the analysis of networks: economies of scale and network externalities.
The first is economies of scale. Consider an industry where consumers want to buy Y units of goods. To keep things simple, imagine that aggregate industry demand Y is inelastic: it does not depend on the average price of the goods. If Y = 10, for instance, consumers really want to buy 10 units of the good in total, although they would rather buy from the cheaper producers. There are N firms competing in the industry. They have the same marginal cost c. The outcome of competition and consumers’ relative price elasticity is a markup m over the marginal cost and thus a price p = (1 + m) × c. Since the firms share the same marginal cost and the same markup, they set the same price. We say that the equilibrium is symmetric. Each firm produces Y / N units and makes a profit mcY / N.
Let us denote by k the cost that any firm needs to pay to enter the industry, and let us assume free entry. Free entry means that new firms will keep entering the industry as long as they expect to recoup their entry cost. Under free entry, we must therefore have mcY / N = k, which means that the number of firms in the industry is given by N = mY × c / k. The term c / k is the ratio of marginal cost over fixed cost. We say that economies of scale are large when fixed costs are large relative to marginal costs, that is, when c / k is small. Large economies of scale therefore imply a low value of N, and a concentrated industry. Note that this is true under free entry, so concentration does not imply rents. It simply reflects the fact that profits must be enough to cover entry costs.
The second concept is that of network externalities. Suppose you belong to a network with h
other people. There is a positive network externality if the value you derive from the network, u(h), is increasing in the number of users, h. Now suppose there are two networks and H people who are thinking about which network to join. Network 1 already has h1 users and network 2 has h2 users. Imagine that you are one of the H − h1 − h2 users who have not made up their mind yet. The values of joining network 1 and 2 are u(h1) and u(h2), respectively. If h1 > h2 and there are positive externalities, you realize that u(h1) > u(h2). You are more likely, then, to join network 1. When your friend makes her decision after you, the numbers are h1 + 1 versus h2. She is even more likely to join network 1. All the remaining users will also choose network 1. Users of network 2 will notice the lack of popularity of their platform and, when given the chance, they will also switch. Absent any countervailing force network 2 will disappear, and network 1 will become a monopoly.
The main way networks can coexist is by differentiating their services. The key idea is to break the simple comparison of h1 and h2 by offering features that are valued differently by different groups of users, or by targeting different income groups. This is how the iOS and Android networks, or the Amex, Visa, and MasterCard networks, can coexist.
Most networks have a mix of direct and indirect externalities. For instance, search engines and GPS systems such as Waze become more precise and more reliable when more people use them.
It is often argued that positive synergies and network externalities are more prevalent today, in the intangible, digital economy, than before. The argument is then loosely used to justify high concentration and dominant positions. I think the argument is misleading, and the case for broad, positive synergies is weaker than most people realize.
The Great Reversal Page 28