I like the intra-generational accounting methodology that Auerbach et al introduce in their paper. By “like,” I mean that I believe it makes a positive contribution to our understanding and should be part of one's toolkit in evaluating U.S. distribution and progressivity issues. But herewith a few comments re. both its limits and particular ways in which it might be used.
Technical measurement issues - Obviously, there are a lot of these, which got considerable attention at both our AM and PM sessions. For example, is capital income actually less tax-burdened than the model assumes? (E.g., due to flow-throughs' ability in many cases to avoid even one level of tax). Should low-earners be assumed to earn a normal rate of return on their saving than high-earners, or to pay higher borrowing rates when they're not entirely liquidity-constrained? The model can of course be adjusted to accommodate alternative assumptions on these and other issues.
What is inequality, and why does it matter? - The paper states that “ultimate inequality” is nothing more or less than “inequality in remaining lifetime spending.” Similarly, it says that “economically relevant inequality” is limited to inequality in spending power. I would tend to view inequality as a more flexible and multi-faceted issue than this language suggests, and I suspect Auerbach agrees.
The paper's implicit model reflects viewing people as deriving utility solely from their own market consumption. The underlying public economics behavioral models would broaden this slightly, treating people as deriving utility solely from their own market consumption plus leisure, but leisure isn't in the measure. If it were, one possible implication would be opening up the question: Are A and B actually the same, if A has $10 million in the bank and lives at the beach, while B has nothing in the bank but has remaining career earnings with an expected value of $10 million? In short, is human capital relevantly the same as wealth? Even staying mainly within the confines of a standard public economics model, I would say this depends on whether B disvalues having to work,
Then there are all the issues around inequality arising from the fact that (as I have discussed elsewhere) the simple public economics model is simply too simple for some purposes. For example, people evidently care about status, prestige, relative position, etc. This not only might increase the social costs of inequality, but also might affect how one needs to measure it. Suppose, for example, that – in keeping, say, with the research by Wilkinson and Pickett – inequality tends to increase stress-related social gradient ills from the top to the bottom, albeit especially at the bottom. Does the inequality that people observe or feel, and that leads to these effects, necessarily track inequality in remaining lifetime spending? Not necessarily. The logic behind the paper's definition is rooted in rational choice assumptions about a particular individual, which (even apart from its incomplete accuracy, especially with the constrained utility function) may be significantly different.
Likewise, suppose that the harms resulting from extreme high-end inequality include its political economy effects. These might range from outright plutocratic capture of the political system to reduced democratic responsiveness, loss of social solidarity, etc. It's not necessarily clear that this, either, would have to track inequality in remaining lifetime spending power better than some other metric.
But again, this just counsels having multiple ideas in mind and using multiple tools – it does not discredit intra-generational accounting as a useful method.
One final point concerns how intra-generational accounting might reasonably be used. Suppose we are evaluating how Paul Ryan's fiscal roadmap would affect the distribution of remaining lifetime spending power. My guess is that – starting with a statiic analysis; I'll note the dynamic issues shortly – it would have a very large effect, especially if the true plan were set forth a bit more forthrightly (e.g., with respect to long-term fiscal sustainability). The plan's elements, so considered, would include enormous tax cuts at the top, and significant cuts (over time, likely far greater than are being acknowledged) to at least the growth rates of privatized Social Security and Medicare, block-granted Medicaid, etcetera.
This might have very large effects indeed on inequality in remaining lifetime spending power. A Saez-Zucman-style wealth measure (assuming it was projected forward, rather than just being computed looking backwards) would fail to show the impact as meaningfully. So would typical tax distribution tables. So the ability to use intra-generational accounting towards measuring the distributional effects of major long-term government fiscal policy changes strikes me as potentially a huge contribution.
But it would require addressing two very large measurement problems. The first is what to do about the fiscal gap – i.e., the fact that current fiscal policy appears not to be sustainable over the long term, and any plans that Ryan has ever announced (disregarding magic asterisks) would make the sustainability problem far worse. Generational accounting tried to deal with this issue by assigning the entire cost of meeting the fiscal gap to “future generations” – a solution that proved confusing (since it was a measurement convention, rather than an actual prediction) and that undermined its acceptance. Not sure what best to do re. intra-generational accounting: use alternative scenarios involving multiple age cohorts?
Second, one has to consider the dynamic issues when designing such a measure. E.g., Ryan et al claim huge positive growth effects, which others believe would be lower even without the “fiscal overhang” issue of failing to fund all of the tax cuts. Again, I suppose one could use alternative scenarios here, including with and without rising interim fiscal overhang.
These difficulties may be too great to support much optimism that intra-generational accounting can play a large part (or perhaps any part) in the political process regarding major policy change proposals. But it may have analytical value for those who are interested in better understanding the distributional issues that are posed by a given policy scenario.