Marginal tax rates


There has been lots of talk about marginal tax rates recently, thanks to Alexandria Ocasio-Cortez’s proposal for 70% marginal tax rates. The essential argument that Ocasio-Cortez proposes is that because these are extremely marginal–for the ultrawealthy–there should be minimal negative impacts from an incidence perspective.

Not so fast, writes Stanford GSB’s Charles I. Jones:

This paper considers the taxation of top incomes when the following conditions apply: (i) new ideas drive economic growth, (ii) the reward for creating a successful innovation is a top income, and (iii) innovation cannot be perfectly targeted by a separate research subsidy — think about the business methods of Walmart, the creation of Uber, or the “idea” of These conditions lead to a new term in the Saez (2001) formula for the optimal top tax rate: by slowing the creation of the new ideas that drive aggregate GDP, top income taxation reduces everyone’s income, not just the income at the top. When the creation of ideas is the ultimate source of economic growth, this force sharply constrains both revenue-maximizing and welfare-maximizing top tax rates. For example, for extreme parameter values, maximizing the welfare of the middle class requires a negative top tax rate: the higher income that results from the subsidy to innovation more than makes up for the lost redistribution. More generally, the calibrated model suggests that incorporating ideas and economic growth cuts the optimal top marginal tax rate substantially relative to the basic Saez calculation.

(Slides are also available.)

This is a super interesting paper. We generally think of the “mega-rich” in an abstract sense as multi-generational heirs and heiresses who did not do much to earn their wealth, or of CEOs who get paid a lot of money for being at the top but would probably not stop being CEO if they were paid less. The first category is actually not particularly true–68% of the ultra-high net worth population is self-made (probably because 70% of wealthy families lose their inheritance by their second generation)–but the second one… may be?

But the really important question is what kind of CEOs these UHNW people are. Do they just hold pencils and push paper and generally keep the trains running on time? Or, to be less extreme, do they spend their time doing procurement and hiring effective salespeople and otherwise doing things that are generally zero-sum and do not broadly add to GDP or social welfare?

But what if they are formerly aspiring CEOs? What if they are the innovators and the scrappers and the upward-mobility-creators of the world? Of course, innovators are, to some extent, doing it for the love of the game–because they see a problem, they think they can fix it, and, sometimes, because of social good (or at least because they can convince themselves they are creating social good). But, one presumes, at least some of this is driven by a desire for super-high payoffs. 90% of startups fail. On a risk-adjusted basis, the only way to compensate for the blood, sweat, tears, and high likelihood of failure, a startup founder may justifiably demand a particularly high amount in order to be willing to do the job. And if this is the case, reducing rewards that a future CEO might expect–for example, through a higher marginal tax rate–might reduce the optimal top tax rate.

That is the exact model that Jones presents and I have to admit that I find it somewhat interesting. It is not obvious that it is true, of course. For one, it is theoretical, not empirical, and it starts by assuming that the previous paragraph is true. For another, it is insanely hard to figure out the social benefit of economic innovation, especially once you start separating by industry (14% of the UNHW population made their money in finance), and it is therefore very hard to price what the top tax rate ought to be. But the intuitive argument of “taxing the top shouldn’t matter because it’s all marginal and on the ultra-rich” is a bit like saying “pharmaceutical drugs are cheap to manufacture so their prices should be low”–it only looks at operating expenses, not capital (and intangible) upfront costs.

Blockchain & Occam’s razor


Matt Higginson, Marie-Claude Nadeau, and Kausik Rajgopal write:

Given the lack of convincing at-scale use cases and the industry’s seemingly becalmed position in the industry lifecycle, there are reasonable questions to ask about blockchain’s future. Is it really going to revolutionize transaction processing and lead to material cost reductions and efficiency gains? Are there benefits to be accrued that justify the changes required in market infrastructure and data governance? Or is a secure distributed ledger primarily just one option when contemplating possible replacements for legacy infrastructure?

Certainly, there is a growing sense that blockchain is a poorly understood (and somewhat clunky) solution in search of a problem. The perspective is exacerbated by short-term expense pressures, cultural resistance in some quarters (blockchains may threaten jobs), and concern over disruption to healthy revenue streams. There are challenges in respect of governance—making decisions in a decentralized environment is never easy, especially when accountability is equally decentralized. And there are technical impediments, for example in respect to blockchains’ data storage capacity.


An emerging perspective is that the application of blockchain can be most valuable when it democratizes data access, enables collaboration, and solves specific pain points. Certainly, it brings benefits where it shifts ownership from corporations to consumers, sharing “proof” of supply-chain provenance more vertically, and enabling transparency and automation. Our suspicion is that it will be these species of uses cases, rather than those in financial services, that will eventually demonstrate the most value.

There are really two starkly different camps in the blockchain debate. On the one side are the unrestrained proponents, who tend to either not understand blockchain as a technology or do not understand the end markets in which blockchain is being deployed (e.g., how screwed up bank back-office technology and workflows really are). On the other side are the unrestrained skeptics, who tend to believe that all of blockchain is hype (in large part, I suspect, because the proponents use blockchain for pretty silly use cases).

Against all this, Higginson et al. propose three conditions to evaluate the usefulness of blockchain:

  • Start with a problem (Occam’s razor)
  • Have a clear business case and target ROI
  • Commit to a path to adoption

It is not as sexy as “blockchain solves everything” and does not access the self-righteous snobbery that comes with “blockchain solves nothing,” but it seems like a pretty reasonable middle ground.

The seven deadly sins of AI predictions


The seven deadly sins of AI predictions