And once they are accounted for, measured wealth inequality plummets.
One reason that the illiquid and non-market resources of DB [defined benefit] pensions and Social Security are typically excluded from studies of wealth concentration is that they are not directly available in household-level survey data. Our work addresses this issue by taking data from the Survey of Consumer Finances (SCF), estimating work histories to predict future Social Security income streams, and combining these results with estimated accrued DB assets and other market wealth holdings to form an expanded wealth measure. We look at households aged 40 to 59, who are building up to peak wealth accumulation before drawing down assets in retirement. Our estimates show that the value of DB pensions and Social Security are [sic] significant relative to other forms of wealth—throughout the wealth distribution but especially at the lower half of the wealth distribution. Indeed, we find that, even for the median household, the present value of DB pensions and Social Security benefits accounts for more than half of all wealth. With respect to their effects on the distribution of wealth, we find that (1) including DB pension and Social Security wealth results in markedly lower measures of wealth concentration, and (2) trends toward higher wealth inequality over time, while moderated, are still present. In particular, the “90/50 ratio”—the ratio of wealth held by those at the 90th percentile of wealth to those at the 50th percentile—is reduced by nearly half for the 50– 59 age group (from 13.4 to 6.8 in 2019) and for the 40–49 age group (10.7 to 6.4) when we include the estimated value of Social Security. The “50/10 ratio” declines even more with the inclusion of Social Security; for 2019, the ratio falls from 13.1 to 4.3 among those aged 40 to 49 and from 21.3 to 4.2 for the 50–59 age group. The share of wealth held by the “top 5 percent” drops from about 72 percent down to 51 percent when defined contribution (DC) plan and DB pension wealth are added to non-retirement wealth; it falls even further, to 45 percent, when Social Security benefits for those aged 40 to 59 are included. The inclusion of each measure, however, has a somewhat different effect: Social Security decreases wealth concentration “at the top,” whether we look at the top 5 percent’s share of wealth or the 90/50 ratio; DB decreases the top 5 percent’s share, but in more recent years, it actually increases the 90/50 ratio. The top 5 percent’s share of our expanded wealth measure rises 8 fewer percentage points compared with the top 5 percent’s share of non-retirement wealth over the 1989–2019 period.
This is from Lindsay Jacobs, Elizabeth Llanes, Kevin Moore, Jeffrey Thompson, and Alice Henriques Volz, “Wealth Concentration in the United States Using an Expanded Measure of Net Worth,” Federal Reserve Bank of Boston, Working Papers, April 2021. HT2 Tyler Cowen and Scott Lincicome.
Economists have known for a long time that something like their claims had to be true. For very wealthy people, defined benefit pensions and Social Security are a small percent of their wealth. For the non-wealthy, defined benefit pensions and Social Security are a large percent of their wealth. So leaving those two things out badly distorts two things: (1) the measure of wealth for the non-wealthy and (2) the extent of wealth inequality.
NOTE: The title is an attempt at humor. One of the early articles I read when I was learning economics in the early 1970s was Robert J. Gordon, “$45 Billion of U.S. Private Investment Has Been Mislaid,” American Economic Review, June 1969.
READER COMMENTS
Kevin Dick
Jul 11 2021 at 9:13pm
I posted this same study to my Facebook feed. Great minds!
An additional point I made was that taking SS benefits into account is crucial if you want to have a coherent discussion of redistribution policy.
You can’t logically argue that we need _more_ redistribution without taking into account _existing_ redistribution.
This same issue crops up in income inequality discussions. People will argue for new redistribution programs based on before tax, before transfer income inequality. But if you use this measure, you will never take into account the effect of such programs and no amount of redistribution could ever possibly be enough!
John Hall
Jul 11 2021 at 9:21pm
It’s a great and obvious point.
Jon Murphy
Jul 11 2021 at 9:27pm
Good point. Just about every discussion of policy is this way. I can’t tell you how many times I see people arguing for things like carbon taxes without taking into account the myriad of mitigation measures already in place.
Dylan
Jul 12 2021 at 8:26am
I completely agree with Kevin on people not taking current redistribution into effect (partly because of limitations in the data as pointed out by this paper). I’ve not seen the same thing in discussions of carbon emissions (with the exception of some explicit, “no mitigation” models). Do you have an example?
Jon Murphy
Jul 12 2021 at 2:02pm
Sorry, I thought I replied earlier. I guess I did not.
Pretty much every discussion of carbon taxes start from a “zero mitigation” assumption. It’s typically not explicit, but it’s implicit. The argument goes something like this:
First: Carbon emissions exist
Second: These emissions fall upon third parties, and therefore are externalities
Third: Therefore, we need a carbon tax to internalize the cost.
The problem is with the second assumption. When the claim is made the carbon is an externality, then they are implicitly claiming that there is no mitigation in effect. Ask just about any economist what the solution to CO2 emissions are and they will say “carbon tax.” Indeed, I cannot think of a single example of a paper or op-ed discussing optimal taxation which takes into account already existing carbon mitigation efforts such as gas taxes, parking fees, auto registration fees, public transportation subsidies, etc. Once we take into account these factors, then there may not be any externality at all; all the costs may already be internalized.
Of course, it’s possible (perhaps even probable) that even taking into account all the current mitigation efforts, a carbon tax would still be optimal. But the conversation doesn’t usually happen.
Dylan
Jul 12 2021 at 2:38pm
I think we’ve covered this before, but I feel I must be missing something. Any model of carbon emissions starts from the baseline of what we’re actually emitting. And those numbers, by necessity, include all of those existing policies that dampen demand. How do we get from there to the implicit claim that there are no mitigation measures in effect vs. the existing measures are not sufficient to fully internalize the cost?
And, a more empirical claim, but I’d expect that the subsidies to the car industry (things like mandated number of parking spaces for any new construction, road construction that is not fully funded out of gas taxes, etc…) more than outweigh the mitigation effects of the items you list.
Jon Murphy
Jul 12 2021 at 3:00pm
Yes, but the assumption is that those levels of emissions are a “free market” outcome.
Jon Murphy
Jul 12 2021 at 3:04pm
Dylan:
Why don’t you drop me an email (jmurph23@gmu.edu)? I think it’ll be a little easier to explain what I mean that way
Dylan
Jul 12 2021 at 4:33pm
My, perhaps naive, understanding of the way this works is 1) Add up all the emissions in a “business as usual” case. Meaning no carbon tax or other new mitigation strategies, but taking into account the policies currently in place. 2) Figure what that computes to in terms of amount of warming. 3) try and put a dollar value to the net cost of that warming. 4) That then helps inform your cost of the ideal carbon tax.
More complicated than that of course, but the basic idea is that the existing measures should already be “priced in.” I understand of course that the models are a far ways away from perfect, tons of unknowns still, lots of assumptions, etc… But am I missing something else basic in my understanding?
Kevin Dick
Jul 12 2021 at 4:43pm
While this is a bit off topic, I would like to observe that I have made the observation on this very blog that a carbon tax should be viewed as a replacement for all less efficient measures and believe I used fuel excise taxes as an example of the issue (I pointed out that an efficient carbon tax wouldn’t actually increase the price of gasoline by very much).
I am quite careful when I advocate for a carbon tax to explain that my support is contingent upon repealing all the other measures.
I’ve read a fair amount of this literature and I think it’s fair to say that many, though far from all, economists are also careful to observe in their formal writing that a carbon tax is a replacement for existing policies.
Mark Brady
Jul 11 2021 at 11:43pm
Okay, but if you make that argument, how would you avoid supporting Social Security and Medicare right now for the presently retired and soon-to-be retirees?
Matthias
Jul 12 2021 at 8:07am
Depends on what you don’t like about Social Security.
For example, you could argue it should be replaced by a system modelled after Singapore’s Central Providence Fund.
JFA
Jul 12 2021 at 8:41am
You don’t necessarily have to favor SS and Medicare to use this data in the current debates on inequality. One just needs to point out that if you are trying to measure wealth inequality, all assets need to be accounted for. You can debate how secure of an asset a future SS check is (though many proponents of the woes of wealth inequality would not want to start making that argument as they often also argue for robust state-sponsored retirement benefits), but by including them, the conversation actually progresses.
This is akin to the exercise of using after-tax-and-benefit income to calculate poverty measures.
David Henderson
Jul 12 2021 at 9:53am
Mark,
JFA has an excellent answer to your question. I think you’re confusing positive and normative.
Let me take another example. Let’s say that a company uses eminent domain to take someone’s house at a below-market price. That house then becomes part of that company’s wealth. To recognize that is not to say that the taking was justified.
David Henderson
Jul 12 2021 at 10:29am
Mark,
An even closer analogy than my eminent domain one: the government gives people food stamps, housing subsidies, and other forms of welfare. As Kevin Dick has pointed out above, these should be counted in people’s income. That says nothing about whether these programs should exist.
Jens
Jul 12 2021 at 4:16am
This seems to show how important these “illiquid and non-market resources” are for a large part of the population. I just don’t really know who to pity for this whole constellation.
Alan Goldhammer
Jul 12 2021 at 10:51am
I just skimmed the paper and will have to give it a more thorough read. One comment that I have is DB plans are disappearing rapidly. My former employer froze its plan back in 2012 which did not impact me as I had already retired. It was a simple DB plan with no inflation adjustment at all; the yearly payment stays the same. Going forward all employees have to rely on their 401(k) for retirement purposes (I can’t be sure but I think the firm has upped the employer contribution to it).
I don’t know if the analysis covers those DB plans that end up insolvent because of financial engineering following takeovers. Such pensions end up cut buy 2/3 when the get turned over to the Pension Benefit Guarantee Association and this is a more frequent occurrence.
Mark Z
Jul 12 2021 at 11:39am
Natasha Sarin and Catherine Sylvain had a paper last year that I think deserves to be mentioned in this context, (link) where they find that “top wealth shares have not increased in the last three decades when Social Security is properly accounted for.” As I understand it, this is because low interest rates have driven up asset prices, but because most of rich people’s assets (stocks, etc.) are counted as assets, and most of poor people’s assets (S.S. benefits) are not, rising asset prices don’t appear to ‘lift all boats.’ But if S.S. benefits were instead privately owned assets, their values would go up as well with asset prices. Not that they’re arguing that people would get better returns if S.S. were privatized or something, but rather that, as an accounting artifact, because S.S. isn’t counted as an asset, low interest rates make inequality seem to increase more than it is really increasing, I think.
Mark Z
Jul 12 2021 at 10:07pm
*Sylvain Catherine, not Catherine Sylvain. Oops.
Dave Smith
Jul 12 2021 at 11:44am
Perhaps quite a bit of wealth tied up in Medicare as well.
David Seltzer
Jul 12 2021 at 11:58am
“we find that, even for the median household, the present value of DB pensions and Social Security benefits accounts for more than half of all wealth.” This is the salient point. I calculate my net worth monthly. I include the PV of both my social security income direct deposits and deferred income. I adjust the discount factor as a function of my age. The time horizon n, is taken from actuary tables. In both cases I’m quite conservative.
Gene
Jul 12 2021 at 2:14pm
David, could you share the formulas/resources you use for these calculations?
David Seltzer
Jul 12 2021 at 4:17pm
Gene, I use an ordinary annuity formula and solve for the present value of SS and DC periodic payments. To wit. PV = Periodic payment* [1-(1+i)^-n]/ i.
For example: 1000* [1-(1+.05)^20]/.05 = 12,464
The present value of an ordinary annuity paying $1000 per year for 20 years discounted at 5 % is $12,464. I used Social Security Actuarial Life tables to estimate my remaining life expectancy and the marginal probability of death to make these calculations. I hope this helps.
David Seltzer
Jul 12 2021 at 5:12pm
Gene, correction: For example: 1000* [1-(1+.05)^20]/.05 = 12,464
should be 1000*[1-(1+.05)^–20]/.05 = 12,464 Apologies.
Gene
Jul 12 2021 at 7:19pm
Thanks David!
Billt
Jul 12 2021 at 1:33pm
Isn’t Social Security underfunded? I always hear about the coming bankruptcy of Social Security. Arnold Kling says that Social Security is like putting an IOU in a jar every payday, then spending all the money you earned. If someone asks about retirement you point to the jar.
The same could be said for many pension systems, California’s for example. Shouldn’t any accounting of individual wealth take into account this underfunding?
Christophe Biocca
Jul 13 2021 at 8:59am
Social Security suffers from two different kinds of underfunding risk:
It does not charge premiums sufficient to cover what it will have to pay out given the returns it makes, making it dependent on a favorable worker-to-retiree ratio to paper things over.
All of its assets are essentially T-Bills (a special kind thereof, but close enough for our purposes).
If you pretend, for the sake of argument, that social security’s obligations and powers are set in stone, the first kind of underfunding is serious (resulting in a ~21% haircut for whoever will be drawing payments in 2035), but the second is not (the government will simply have to raise taxes or borrow more from other parties, to make up the shortfall).
The more realistic treatment would be to treat social security as the political program it really is, but the problem with that is that might work to the retirees’ favor (they’re a big voting block) or against them (other spending may take priority) and it’s impossible to tell in advance which one will happen. For example if Roth 401ks lose their “not taxed on the way out” tax treatment and that money is used to fix the social security shortfall, or if Social Security benefits become means-tested, that’ll reduce wealth inequality even further than the analysis suggests. If the US reneges on all its debt (including SS IOUs) or hyperinflates its currency (and removes the cost of living adjustment for benefits), then it may very well increase.
Thomas Lee Hutcheson
Jul 12 2021 at 1:47pm
Obviously the number you get depends on the definition of assets. To know what it means, one would like to know how this measure has changed compared to changes in the w/o SS and Medicare.
Comments are closed.