(pdf)
Though the cost of the COVID-19 pandemic has been enormous, a silver lining can be found in the fact that it arrived in the U.S. in 2020 and not thirty years earlier before internet-based services revolutionized our ability to connect with one another. The incredible pace of innovation over that time allowed many Americans to continue working out of the office, see the faces of loved ones they couldn’t see in person, and receive virtual medical care.
Much of that innovation has been driven by an American tech sector that is now under assault from politicians of all stripes and at all levels of government. Unfortunately, rapid change often inspires elected officials and regulators to explore entirely new bad ideas. One such idea is that of a “data tax,” sometimes referred to as a data dividend.
In their first iterations, data dividend proposals aimed to extract dollars from large technology companies in order to produce some form of payment to individual users. This is rooted in the premise that data is akin to commodities with inherent value, like metals or minerals, and thus economic benefits derived from its use should flow to the individual that “owns” it.
However, that idea quickly morphed into something very different (and yet very familiar): a new tax that would generate millions in revenue for cash-hungry governments. These modified proposals would institute huge new tax burdens on top of existing business income taxation, with gargantuan administrative complexity to boot. Because they use the tax code to achieve aims unrelated to revenue generation for necessary functions of government, these expensive and burdensome data tax proposals are dangerous developments in state tax policy.
What is a Data Dividend?
The premise of a data dividend is that companies are simply harvesting data from consumers who don’t know how valuable their personal information is, then profiting off of something they rightfully should have paid for. Thus, the data dividend aims to rectify this perceived imbalance by imposing a new tax (on top of existing business income tax obligations) and redistributing those funds.
One of the first major examples of such a proposal came from California Governor Gavin Newsom. The exact form Newsom’s proposal would take remains somewhat nebulous, but it would likely be some version of a tax on the revenues of companies that utilize consumer data, the revenues from which would then be passed on to Californians via direct support or more indirect aid like so-called “baby bonds.”
Governor Newsom is not the only politician to come up with a data dividend proposal. Former Presidential candidate and current New York City mayoral candidate Andrew Yang, for example, helped found the Data Dividend Project (DDP), an initiative which aims to collectively bargain for users to get paid for their data. Like many data privacy advocates, the DDP appears to vastly overestimate how much users could receive for their data, with Yang anticipating “something like $20, $50, or $100.”
Another proposal out of New York gets into slightly more detail. Assembly Bill 9112, introduced by Assemblywoman Stacey Pheffer Amato, would impose a five percent gross receipts tax on “every corporation which derives income from the data individuals of [the] state share with such corporations.” Unfortunately, this recklessly vague language would mean new tax burdens for a wide variety of businesses, including many that have no relation to the Big Tech companies that proponents seek to target. Businesses with rewards cards, companies offering free WiFi, auto insurers, and even restaurants could see themselves saddled with new taxes if this bill were enacted.
Finally, another New York product reaches the logical endgame of progressive efforts to tax digital companies’ profits. Instead of creating some fund through which individuals would be paid for their data, more recent proposals have been straightforward tax increases to fund new government programs. Thus, New York State Senator Andrew Gounardes introduced a proposal not to return revenues raised from data taxes to taxpayers, but instead to use the revenue to “invest in digital literacy and workforce redevelopment” in New York. Gounardes’s proposal is supported by New York City Mayor-Elect Eric Adams.
It’s important to distinguish these efforts from taxes on the use of data like the California and New York proposals described above from taxes on the sale of data. The latter is a far narrower tax, and one which would largely not affect social media and internet giants. Legislators in Washington and Oregon, for example, have proposed gross receipts taxes on the sale of personal information. These taxes would specifically target data brokers, or businesses that profit off of direct sales of consumer data. Businesses that simply collect consumer data in the course of operations, even if they use it to make their platforms more profitable, would not be affected, so long as they do not sell the data itself.1
There is one key problem (among many others): it’s simply not true that consumers receive no benefit for the use of their data.
For example, social media platforms are often targets of privacy hawks for their use of consumer data. But users of social media networks like Facebook and Twitter don’t pay a cent for a platform that allows them to connect with people across the world. The reason why this is the case is that social media platforms are able to generate profits outside of the ordinary structure of pricing access to the platform for individual users. Instead, they price access to the platform for advertisers, who pay for the privilege of highly-targeted, cost effective advertising for their businesses.
Unfortunately, states have also been pushing hard to tax digital advertising as well, even when they do not tax traditional advertising.
Just Like the Alaska Permanent Fund?
Advocates of a data dividend often invoke the Alaska Permanent Fund (APF) as a model for how a data dividend could work. After all, as people love to claim, “data is the new oil!” This, however, is a bad comparison.
The APF is an investment fund that receives 25 percent of all royalties from development of the state’s oil, gas, and mineral resources. Annual dividends are then paid out equally to all Alaskans, usually totalling between $1,000 and $2,000.
The first and most obvious difference between the APF and “data dividend” proposals is that the APF was a reallocation of existing government revenues, not an effort to raise new ones. The APF was born out of taking existing royalties and investing them, not creating an entirely new tax on drilling or mining. Companies paying those royalties would care little whether they went into a general revenue fund or a separate one to be paid out to state residents, and it would have little effect on their behavior.
On the other hand, a “data dividend” would cost the paying businesses as well as the consumers receiving it. Tech companies would be forced to pay a new tax, likely costing consumers access to the services they valued.
Another difference is one of scale. The APF has proven resilient in part because it has succeeded in paying out substantial checks to Alaskans on an annual basis. Californians, New Yorkers, or residents of any other state would be far less enthusiastic about an annual check that totals a few cents, as a data dividend likely would.
Taxpayers receiving directly some portion of the royalties from harvesting of their state’s natural resources is a very different scenario than a misguided effort to help taxpayers “benefit” from services they already get to use for free. The two should not be mistaken.
Tax Policy as a Weapon
Taken together, these proposals amount to a push to raise revenue off of profitable tech companies with little regard for how that will affect a business model that is enormously beneficial to consumers, and with little regard for what constitutes appropriate tax policy. Saying “I can’t believe this website is free” is such a common Twitter meme that the company itself even got in on it, but what if it weren’t? Take away the tools such companies have to turn a profit, and they’d be left with little choice.
Would that be better for consumers? Hardly. While many Americans profess concern over how their data is used by businesses, this is a case where revealed preferences contradict stated ones. After all, keeping Facebook, Twitter, and Google from collecting your data is as simple as ceasing to use the services they offer, yet Americans continue to do so in droves.
After all, a swap between paying for the use of websites with one’s data and paying with actual money is a bad trade for consumers. A consumer’s data is worth next to nothing to that consumer — you can’t buy a pizza by offering the restaurant your preference for Fleetwood Mac. And as far as Americans value their data privacy, as previously stated, it is clearly less than they value the use of the services their data pays for.
At the same time, it’s not like users are simply trapped — some 15 million Americans quit Facebook in the wake of the Cambridge Analytica story. In short, those Americans who valued their privacy more than the use of Facebook left, and those who did not stayed. And that’s exactly how it should be.
It’s also important to note that most major companies, such as Facebook and Google, do not “sell” data in the traditional sense. Instead, they use it to offer better targeted advertising to prospective advertisers. Advertisers do not actually receive access to the data that platforms collect. Rather, they indirectly benefit from it by virtue of being able to take advantage of the platform’s ability to put their advertisements in front of the users they want to reach. This fact also makes valuation of data very difficult — how much of the value of a platform’s advertising space is derived from its efforts to organize and extract useful information for ad targeting purposes, and how much is derived from the data inputs into that system in the first place?
One independent study came to a pretty stark conclusion: an individual consumer’s data is worth little, even if you assume the platforms themselves add no value. The Electronic Frontier Foundation estimated that Facebook’s revenues amounted to $7 per user in 2019, the vast majority of which came from advertising. Their profits after accounting for costs, of course, were far lower.
Even if Facebook split the profits from data sales evenly with consumers, consumers would be receiving checks worth a matter of cents. Even that may be overselling it — when an entrepreneurial journalist attempted to get paid for his data back in 2018, he ended up raising a grand total of 0.3 cents.
Ultimately, these proposals are simply the next in a long line of efforts to use the tax code to achieve aims other than generating revenue to fund necessary functions of government. The Tax Foundation articulates several key principles that constitute sound tax policy, including simplicity, transparency, neutrality, and stability. Data dividend proposals violate each one of them.
A data dividend would be one of the most complex and burdensome taxes administratively if one were enacted. The enormous complexity of the data economy makes it difficult to effectively target and structure such a system. A broad-based tax that applies to any company using data in its operations would encompass nearly every business in existence, imposing burdens that even data dividend supporters would likely find untenable. Narrowing the scope to effectively exempt restaurants and companies with rewards cards, among others, would require precise definitions of what activity is covered by such a tax, in an area of business that does not lend itself to such precision.
Data dividends also fail the test of transparency, not least of which is because they rely on the blunt and destructive tool of gross receipts taxes to achieve their aims. Gross receipts taxes are notorious for causing issues with “tax pyramiding,” and also suffer from the structural flaw of making no account for profit margins when imposing tax. This means that big established businesses with high margins will face lower effective tax rates on profits than smaller businesses with narrower margins, some of which could actually see a gross receipts tax swallow up their profit entirely.
There is no test a data dividend fails more comprehensively than that of neutrality. Tax policy should not be used to benefit or hinder certain types of businesses or activities. Instead, it should be used to generate revenue in an efficient manner to help fund limited, necessary government functions. Data dividends, on the other hand, exist precisely to hinder certain types of business activity, largely by layering new taxes on top of existing business income tax regimes that affected companies are already complying with.
Finally, with regard to stability, data dividends fail once more. Stability is not served by the creation of novel tax regimes to target businesses that are disfavored for reasons unrelated to tax policy. Furthermore, many data dividend proposals suffer from potential legal issues that could see them tied up in court for years in legal limbo, not unlike their close cousins in digital advertising taxes. This could put millions of dollars in revenue in serious doubt.
Conclusion
As long as the “techlash” continues to be a reality in American politics, policy options that appear to hurt tech companies will doubtless be tempting to politicians. Yet spite is a poor justification for policymaking, and “data dividend” proposals are a classic case of cutting off the nose to spite the face.
Taxpayers are in no way served by losing access to free services, and tax policy is in no way served by using discredited tax structures to impose new layers of taxation on top of traditional business income taxes. Policymakers should avoid trying to capitalize on public sentiment in a way that would only harm the people they claim to help.
1 While ostensibly related, taxes on the sale of data implicate other elements of tax law and are deserving of separate study and analysis which we are not attempting here.