Tuesday, February 07, 2017

NYU Tax Policy Colloquium, week 3: Alan Auerbach (et al)'s "U.S. Inequality, Fiscal Progressivity, and Work Disincentives: An Intragenerational Accounting" - Part 1

Yesterday my frequent past colloquium co-convenor Alan Auerbach came to the NYU Tax Policy Colloquium to discuss the DBCFT how we should measure the fiscal system's progressivity. His paper (coauthored with Laurence Kotlikoff and Darryl Koehler) draws on earlier work (with Kotlikoff) that I've always liked and considered important, despite criticisms that it has received elsewhere, concerning generational accounting (GA).  GA aims to look at how the fiscal system - taxes, transfers, and whatever spending one can reasonably classify as funding consumption by particular individuals - affects generational distribution. More shortly on a couple of the issues that raised. The current paper, as per its title, engages in "intragenerational accounting" - measuring the fiscal system's distributional effects within an age cohort, in particular as between richer and poorer individuals (although a key part of the issue is how to classify people in this vertical sense.

I'll discuss the paper in two separate blog posts, so as to keep this one from getting too long.

To place the paper's analysis in context, consider the following chart from Table 1 of a 2014 NBER working paper by Auerbach's Berkeley colleagues Emmanuel Saez and Gabriel Zucman (containing findings that are also mentioned in their forthcoming Quarterly Journal of Economics Paper):

Wealth group
# of families
$$ threshold
Average wealth
Wealth share
Full population

Top 10%
Top 1%
Top 0.1%
Top 0.01%
Bottom 90%


One point from this table that has received a lot of attention is that the top 0.1% in the U.S. population, in terms of families ranked by wealth, holds almost as high a share as the bottom 90%. A second widely noted point, also mentioned in Saez-Zucman 2016, is that the wealth share of the top 0.1% has more than tripled since 1978.

But how should we think of wealth as a measure, in assessing U.S. progressivity and distribution? Auerbach et al note in their paper that its defects include the following:

1) It doesn't include the value of human capital, i.e. the present value of one's future earning power. To show why this might be important, suppose there are 2 people, one with $10 million in the bank and no job (or plans to get one), and the other with a job and/or career, as secure as the first one's wealth, that is expected to generate earnings with a present value of $10 million. These two individuals seemingly can afford the same value of remaining lifetime consumption (including bequests, which Auerbach et al note can often be thought of as a personal consumption choice, preferred to other uses by the individual who leaves the bequest). A big difference, of course, is that the first one doesn't have to work, while the second one does. But - suppose the second individual actually likes working, and indeed prefers it to "leisure."

2) It doesn't include the net positive value (if any) of expected future net transfers (minus taxes) from the government. Now, while Social Security obviously is not generous enough to offer anyone expected future benefits with a present value of $10 million, if it did (and if these benefits were politically secure) that person - also leaving aside liquidity issues - would be able to afford just as much remaining lifetime consumption as the two whom I mentioned above. Now obviously, Social Security and Medicare "wealth" (which raises further issues, since it has to be consumed in the form of healthcare) isn't going to do much about the wealth numbers we observe at the top. But it would discernibly raise numbers at the lower wealth echelons.

3) It's a snapshot that doesn't count for lifecycle effects. In a standard pattern, people have little wealth (unless they have inherited it) when they are young, then gradually build up some wealth as they save during their working careers, then spend it down during their retirement years except insofar as they are leaving bequests (either deliberately or as a byproduct of precautionary saving).

Suppose - although we of course know that this isn't the case - that the Saez-Zucman wealth charts were solely a function of lifecycle effects. That is, suppose everyone did identically on a lifetime basis, but started at the bottom when they were young, gradually marched up to the top 0.01 percent as they reached their late peak working years, and then went down again (say, to zero wealth at death) during retirement. Then the Saez-Zucman chart would say nothing whatsoever about wealth distribution in our society, at least as conceived of on a lifetime basis. Each of us would be taking the same full ride. It would still be interesting to know why things have changed so much since 1978, and there might be issues of failed lifetime consumption smoothing - meaning that we might need policies to make it easier for people to borrow or save as needed to consume as much they wanted (given the lifetime budget constraint) in low-wealth periods. But everyone would be doing the same on a lifetime basis.

Now, I am convinced that the Saez-Zucman info is extremely important despite all these issues (and I would guess that Auerbach, whether or not all of his coauthors, agrees). But they offer a system of measurement that aims to adjust for these issues. It aims to group people in a given age cohort by the present value of their remaining lifetime spending power, and also to look at the fiscal system's impact on where they end up.

One finding is that U.S. inequality is far less extreme by this measure than by the Saez-Zucman wealth measure. Now, surely we already knew that. Saez-Zucman would be portraying a horrific dystopia, featuring mass privation in the U.S. population, if people could only consume - including necessities such as food and shelter - by using saved wealth. Insert Trump Administration joke here if you like, but we know that things aren't actually currently like that here. And it's by reason of the excluded factors, such as earning capacity and U.S. transfer programs (especially during retirement) that things aren't so dire. But looking ahead for a moment, that doesn't mean the Saez-Zucman finding is "wrong' or irrelevant - just that one needs to think further about what its significance might be. The Auerbach et al measure of expected remaining lifetime consumption is also highly relevant, and for some purposes clearly the more informative of the two.

A second finding in the Auerbach et al paper is that the U.S. fiscal system is significantly redistributive. People at the high end, by its metric, appear to pay far higher marginal and tax rates than those lower down. One reason for this is the U.S. transfer system, especially for retirement. A second reason is that we do indeed tax capital income, and thus saving to fund future consumption (or bequests). Indeed, between the entity-level corporate tax and owner-level capital income taxes on saving, the effective rate as measured may exceed current year marginal tax rates.

A third finding is that there's lots of dispersion in marginal tax rates, as among people whom the study groups at the same level vertically. Loss of Medicaid benefits when one places out of it on earning or asset grounds is one reason for this at the lower end. The paper looks at marginal tax rates by examining what one happens to one's taxes and transfers if one earns an extra $1,000, and might come out differently if one were asking, say, the effect of permanently earning an extra, say, $1,000 per year (which in some cases might be a better rendering of the actual marginal choice that a given individual faced).

Auerbach et al haven't done similar studies for other countries, which would require a whole lot of data and also local detail regarding the relevant fiscal systems. But it's likely that our peer countries show significantly greater after-tax-and-transfer equality in remaining lifetime spending power. To make good cross-country comparisons, however, one might need to tweak the treatment of government spending to adjust for the fact that other countries often offer a lot more by way of benefits that are methodologically tricky to assign to particular individuals, and yet that are enjoyed sufficiently pro rata (while being funded through VATs and income taxes under which the rich pay absolutely more) to have a potentially significant bottom line impact.

Further comments on intra-generational accounting to come in Part 2, which I hope to post shortly.

No comments: