Yesterday I wrote that various candidates’ promises to eliminate the Windfall Elimination Provision were, most charitably, misguided, but, more likely, simply pandering, because the WEP is an entirely appropriate provision to prevent certain teachers from “double-dipping” in their retirement benefits. And in response, some commenters (on twitter and my personal website) suggested that while the WEP reductions seemed fair, the GPO, or Government Pension Offset, was unreasonable.

In response, here’s a follow-up: yes, the GPO is also reasonable and prevents double-dipping. But to explain that, I have to address what seems like a wholly different topic first, so be patient!

Among the various promises that the Democratic presidential candidates such as Pete Buttigieg and Amy Klobuchar are making is this: they will institute a system of “caregiver credits.” Buttigieg’s white paper describes his proposal as follows:

“ Those caring full-time for children or for a disabled or elderly family member do not currently receive credit toward Social Security benefits—disproportionately affecting women. Under Pete’s plan, Social Security will finally recognize the undeniable: caregiving is work. Caregivers of a child, elderly, or disabled dependent will be awarded credit toward Social Security benefits as if they earned the median earnings of a full-time, year-round worker outside the home, with no limits on the number of years for which caregivers can claim the credit.”

Are caregiver credits a good idea? Maybe, if implemented judiciously, and I’m not sure the Buttigieg proposal hits the mark, without duration limits and with a structure that boosts benefits for women out of the workforce up to the median level, that is, above, by definition, half of the workforce. There are a variety of ways to provide benefit boosts to mothers: income-splitting credits credit to stay-at-home mothers half their husbands’ income; Canada reduces the number of years included in the averaging period; Switzerland and France increase benefits for parents regardless of their work choices; and, yes, Germany and Austria provide caregiver credits, with, in Germany for example, median-income credit given to caregivers of under-3s, a boost to part-time workers juggling work with caregiving, and partial credits for parents of under-10s.

And, as it turns out, the United States already has a form of caregiver credits; it’s just that it’s a simplistic and old-fashioned system that’s not perceived that way. In part, Social Security uses only 35 years of averaging to reflect some absences from the workforce over one’s adult years. But in addition, a woman who has been out of the workforce for a significant length of time is provided a special form of minimum benefit, in the form of 50% of the benefit earned by her husband. (Yes, the Social Security website uses confusing wording, with a subject header “Benefits For Your Spouse” to describe benefits individuals may be eligible for as a spouse, but so be it.) Of course, this doesn’t benefit out-of-the-workforce parents who aren’t married, or other sorts of caregivers, but at the time when Social Security was designed, this sort of minimum benefit made sense.

Again, with respect to benefits for surviving spouses of deceased Social Security participants, the wording of the website suggests that workers earn these benefits for their spouses, and even calls them a form of “life insurance.” But, again, once the surviving spouse reaches retirement age, the deceased spouse’s benefit serves as a minimum benefit to what she or he has earned in her own right.

The bottom line is that in an employer-benefits based system, workers earn benefits for themselves and their families. They earn health insurances for family coverage. They earn life insurance benefits that pay out to designated beneficiaries in case of their death. And if their employer provides pension benefits, those benefits provide the opportunity to select a “joint and survivor” form in which their spouse receives a specified portion of the benefit after their death — and the latter is true regardless of what income that spouse might have, because it’s a benefit earned by the deceased spouse.

But Social Security isn’t an employer benefit. It’s Social Insurance. And the rules are different. Regardless of how they may be described, the reality is that these are benefits accruing to the recipients themselves, even if they are “earned” by means of being married to an earner. A non-working mother truly receives her own Social Security benefit even if based on her husband’s income record; it is not the case that he receives an extra benefit because he has a dependent wife, and it is not the case that he has a government-provided life insurance policy to the benefit of his wife.

And that gets us, finally, to the Government Pension Offset.

This offset applies to workers in federal, state, or local government who have opted out of participating in our nationwide social insurance system, or whose employer has chosen to do so.

The fundamental calculation is this: the amount of benefit an individual would have been due as a spouse or surviving spouse minimum benefit, is reduced by two-thirds of the value of “opt-out” government pension. If an individual would have been eligible for a $500 minimum benefit as the spouse of a Social Security-covered worker, but had worked for a state pension system that opts out of Social Security and earned a $600 pension, then two-thirds, or $400, would be subtracted from the minimum-benefit as-a-spouse, for a net of $100. (The example comes directly from Social Security.)

The principle is entirely fair: the minimum-benefit-as-a-spouse should only ever be a minimum, not an add-on. To get both the minimum benefit and the full own-benefit is double dipping every bit as much as if this were the case by stacking two Social Security benefits together.

Now, at the same time, in a perfect world, the math would be different.

After all, a state or local pension benefit is a Social Security-replacement benefit and an employer supplemental benefit all wrapped into one. A more precisely-fair calculation would split the state or local pension into two numbers: the portion that replaces Social Security, and the portion that supplements it in the same way as private-sector pensions do. Then only the first of these would be compared to the benefit-as-a-spouse to identify whether Social Security should pay out a minimum benefit.

Now, whether Social Security has the data available to do the math, I can’t say. And even if so, one could make the case that an individual who has participated in an alternative pension system for their career has made a choice to opt out of Social Security in a way that’s not true of a non-earner or low-earner. And for middle/upper income full-career workers and for their employers, opting out is a win, financially, because they opt out of subsidizing the poor with a bend-point formula in favor of earning benefits at a flat rate.

And, again, the solution to this unhappiness is simple (and I cannot stress this enough): those 15 states whose employees do not participate in Social Security should be moved into the system like the rest of us. Why those states — which include notoriously-poorly funded California and Illinois — don’t do so is plain: it would increase their costs and force them to pay up-front in the form of FICA contributions.

(And, incidentally, the federal government itself made the switch back in 1984, at which point newly-hired workers as well as those who chose to make the switch, began to be covered by Social Security.)