Facebook co-founder Chris Hughes has a provocative new essay in The Guardian arguing for “a new tax on the revenues of any companies that collect and store meaningful amounts of information and data about people to build their businesses.” While his proposal is not fully fleshed out, it’s nonetheless interesting to think through. Ultimately, I find myself skeptical of the need for a special tax on data companies, for a few reasons:
1) Operationalizing a “data tax” into law would be very tricky
What counts as “data” is hard to pin down metaphysically, much less as a matter of law. As a result, I suspect any attempt to operationalize a data tax into law is likely to be highly technology- and medium-specific. Is it phone meta-data? Keystrokes? The content of what you post online, or also your lurking behavior? Is it only data that’s permanently stored? Is it just the raw inputs or also the new data created after it has been transformed? What about satellite data tracking the movement of cars to inform a highway project? Do I own my driving patterns? How are all these kinds of data weighed against one another for relative significance or degree of ownership?
2) We shouldn’t tax mobile factors
The rule of thumb in tax policy is to tax the factor of production that is least mobile. Land and other natural resources, wages and consumption are all great candidates. Data, on the other hand, are perhaps most mobile factor of production ever created. A data tax is therefore likely to create a lot of inefficient avoidance behavior, including firm restructuring. Maybe the “data layer” part of the software stack gets off-shored. Or maybe the data tax spurs the creation of third-party data holders who specialize in minimizing tax incidence through creative accounting, and otherwise take-on various novel forms of liability that an explicit ownership stake on one’s data will create.
3) Data creation is a public good
One person’s data is worth little, but the collection of lots of people’s data is what fuels the insights that companies use to make more money or networks, like Facebook, that marketers are so attracted to. Data isn’t the “new oil”, as some have claimed: it isn’t a non-renewable natural resource that comes from a piece of earth that a lucky property owner controls. We have all pitched in to create a new commonwealth of information about ourselves that is bigger than any single participant, and we should all benefit from it.
This is an astute point, but ultimately cuts against the argument for a special data tax. In essence, Hughes is saying data has all the properties of a public good:
- data is non-rival; it can be duplicated ad infinitum;
- data is relatively non-excludable; platforms and data brokers are able to capture comparable data many times over;
- user generated data has large benefits that are hard for any one individual to capture.
All told, this suggests we ought to (if anything) subsidize data creation, not tax it. Which is roughly what we’ve done throughout history: Before VC backed companies began launching micro-satellites into orbit we had GPS systems run by NOAA and DoD; before Facebook knew our entire life story we had social surveys run by the US Census, and identity services provided by the SSA and DMV; the list goes on. Private data creation has thus in some sense beat the odds thanks to the bootstrap provided by bundling data generation with valuable, free services. In other words, overcoming the chicken and egg problem of building a data-centric platform requires capturing the benefits of network effects. Internalizing the network externality with a data tax wouldn’t just lead to marginally less data; it would potentially jeopardize the very pricing strategy two- and multi-sided platforms companies depend on to grow.