Bryan Caplan  

Means-Testing and Political Economy

Is Labor Law More Oppressive f... Means Testing Isn't "Awesome"...
Why not means-test Social Security and Medicare?  On the surface, this seems like a perfect liberaltarian reform.  Libertarians should favor drastic cuts in government spending, liberals should favor drastic cuts in government spending on the rich, and both should favor the brighter fiscal future that the means-testing of major programs implies.

I've converted several libertarians to the gospel of means-testing, but - to the best of my knowledge - zero liberals.  What gives?

The standard liberal objection is that means-testing puts us on the slippery slope to abolition.  Once we means-test Social Security and Medicare, voters will perceive these programs as mere "welfare."  And Americans hate welfare.  To preserve transfers to needy Americans, we have to keep these programs universal.

These fears inspire the following hypothetical.  Suppose the U.S. adopted the following means-testing formula for Social Security and Medicare:

1. Full benefits for everyone below median income.

2. For everyone above median income: For every percentage-point by which your income exceeds the 50th percentile, your benefit goes down by 2 percentage-points.  So if you're at the 53rd percentile, you only get 94% (=100-2*[53-50])% of the full benefit.  If you're at the 90th percentile, you only get 20% (=100-2*[90-50])* of the full benefit.

Notice that a simple back-of-the-envelope calculation says this will cut the cost of Social Security and Medicare by 25%.

My question: In the new political equilibrium, how much do you predict the full Social Security and Medicare benefit will fall?  Remember, this is the benefit that everyone below the median income gets.

I doubt the full benefit would fall by more than 20%.  After all, welfare and Medicaid exist right now.  My long-run point prediction is actually that the full benefit will be larger than it would have been.  The main cause of big transfer cuts is probably fiscal crisis - and my means-testing proposal will push any U.S. fiscal crisis far into the future.

I'm especially curious to hear non-liberarians' predictions about the new political equilibrium.  Please show your work.

COMMENTS (26 to date)
Anon writes:

[Comment removed for supplying false email address. Email the to request restoring this comment. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Kyle Thompson writes:

As a liberal who supports means-testing social security and Medicare, I resent the implication that I don't exist!

Peter writes:

I have been proposing means testing (though not as eloquently and without the math and national stage) for a long time. Social Security and Medicare definitely ought to be means tested. I hear the same basic argument against means testing; that do so will create resentment and different classes of recipients.

I predict the full benefit of SS and Medicare would rise if the programs were means tested. My reasoning is that by making the programs financially sound, it would put the US Gov in a much stronger economic position. A sound, workable US Gov balance sheet would benefit the entire society. Most people do not fully realize just how big a drag on the economy the national debt and future obligations are.

JLV writes:

Consider a world where SS and Medicare are means-tested.

Does Bryan Caplan (and his fellow libertarian means-testing advocates):

a) vociferously argue for the abolition of SS and Medicare


b) vociferously defend the means testing they argued for.

My bet is on a. So why bother with means testing, if we're going to argue about abolition anyway?

Henry writes:

That means-testing "reduces the size of government" reminds me of a criticism of Grover Norquist's no tax increases pledge, which also included removing tax breaks that amounted to de facto subsidies. Is it always better for government to be "smaller" by or does is "smaller government" merely a proxy for some other desirable characteristics?

For instance, imagine a government that taxed 100% of GDP. But it spent money in the exact same way as if there were no taxes at all (except for small administrative costs). So if you earned $100,000 a year, it would all be "taxed", but you could choose for ~$100,000 in "government spending" to be distributed as you want, i.e. you could replicate how your private spending would be in a world without taxes. Would you not prefer to live under this government than the status quo, even though it is ostensibly "large"?

Likewise, universal programs make measured government spending considerably larger than strictly means-tested programs. But if your concerns about large government are that it's "distortionary" and/or "redistributionist", it's not clear which performs better.

Henry writes:
"smaller" by or does is "smaller government"

Should be: "smaller" or is "smaller government"

Frank Howland writes:

I am a liberal with some libertarian views. I agree that means testing would certainly help our fiscal situation but I think that Bryan Caplan understates the slippery slope argument. Cash benefits to poor people are extremely unpopular--that's why TANF is actually quite small relative to non cash benefits like food stamps and Medicaid. The Earned Income Tax Benefit is a larger cash benefit than TANF but is cleverly tied to the income tax and therefore is less visible. (I don't think that the better work incentives that go along with EITC are what make it more politically popular for voters.)

Thus I believe if Social Security were perceived as a cash benefit for poor people it would lose substantial support.

As for Medicaid, not only is it a non-cash benefit, but a substantial part of Medicaid spending goes to nursing home care and much of that to people who are not in poverty. On paper Medicaid is terrific insurance, but because reimbursement levels for Medicaid to hospitals and doctors are very low, it's pretty spotty insurance.

I'd predict that if Medicare were means-tested, reimbursement levels for Medicare to hospitals and doctors would fall toward the very low levels of Medicaid.

John David Galt writes:

The percentage calculations in point two seem to confuse the concepts of average and median. The median income in the US is well below the average. If you did a histogram of income today as frequency vs. dollars, you'd see, not a bell curve but a big hump at around $20k a year, with very few households much below $10k but a long, broad tail that extends well above $200k before it gets narrow. The median is in the neighborhood of $25-30k but the *average* is more like $45-50k.

Still, this phase-out proposal would probably reduce costs by more than 25%, since the people whose benefits it would cut will mostly be those who get the highest benefits now.

SheetWise writes:

How about we come clean, and simply tell everyone who doesn't already know that the program was always a fraud, and that no trust accounts ever existed past the accounting charade. It was always a tax, and your employers did not contribute "their" money. Since the SSA did manage the accounting fraud well, we can easily issue everyone bonds equal to their total contributions, plus interest, less withdrawals. From there, we can start over.

LowcountryJoe writes:

I think it is completely immoral to means-test old-Social Security and take away the opportunity for those who made the most contributions to get nothing. The 'insurance' was designed to provide benefits to those who outlived the mortality table. The only fair thing to do is to substantially raise the age at which people are entitled to full benefits and significantly reduce the benefit for those who take it early.

James A. Donald writes:

A means tested social security approximates the Australian system.

Under the Australian system, most people fund their own retirement, thus Australia, unlike Europe and America, does not have a gigantic off-the-books deficit.

Liberals like to have a gigantic off-the-books deficit because government making promises it cannot and will not fulfill makes government more popular than it otherwise would be.

Similarly, Obamacare and most European healthcare guarantee everyone all the medicine they might ever need. So if an elderly person shows up at hospital with breathing difficulties, instead of giving him treatment, which would result in him occupying a hospital bed for many years, the British National Health give him deep benzodiazepam sedation, which makes his passing more comfortable, and guarantees he passes in a week or so, instead of several years or so, swiftly freeing up his hospital bed.

Australia, on the other hand, does not guarantee all the medicine that everyone might need, so the Australian free healthcare system lets people die who might well be saved, but refrains from murdering them. Liberals naturally prefer a system that guarantees unlimited healthcare, and then murders people when they show up to cash in on the guarantee, because that makes more government more popular.

TimG writes:

What does "means testing" mean in the context of seniors that are generally asset rich and income poor. Shouldn't a libertarian be concerned if the government has enough information about a household's wealth to decide SS and medicare benefits? To me that's a huge advance of governmental power.

Scott Sumner writes:

I think this would be a bad idea. The numbers make no sense unless you are referring to income levels of the retired. Suppose a couple saved a million dollars during their life and turned it into a $50,000 annuity on retirement. Also suppose they earned $50,000 combined from Social Security. If the million dollar nest egg puts them in the top 10% of retirees, they lose almost all their Social Security. That's a huge disincentive to save. And recall that the tax system already has other saving disincentives built in, so it would make the bias against saving much stronger.

Means testing is just another high implicit marginal tax on people who work hard and save.

Telnar writes:

I see two details of the means testing process which strike me as important:

1) Should means testing be based on lifetime income or current income?

My leaning here is to means test a subset of the payment by current income and treat that portion as a welfare payment to insure that those who invested unwisely or unluckily still have something. To address privacy concerns, this benefit would be provided on application only and most taxpayers would not be eligible. The majority of the benefit would be means tested based on lifetime income (using data the IRS already has) to preserve incentives to save.

2) On what time horizon should means testing be phased in?

Here, I don't think that beneficiaries should face a larger short term change than taxpayers as a percentage of per capita program spending/taxes.

MG writes:

I think Bryan's scheme should never be implemented as sketched, for reasons that various posters have alluded to. However, there are variations of the theme that may satisfy those willing to compromise. Andrew Biggs' analysis is the best I have seen.

I still think it all boils down to somw form of government takings. As consolation (compensation) I would prefer that any scheme that reduces promised benefits at least disclose the NPV loss being incurred by the means-tested, either to be worn as a badge of honor or to be compensated for in some form of government equity payable if and when things improved.

MikeDC writes:

I don't understand why anyone would waste time talking about this.

Social Security is already means tested; every dollar of SS benefits is considered taxable income.

Thus, much of the benefits that accrue to rich beneficiaries is already recovered by the government. What's the point of arguing about the accounting mechanics of it, which are almost entirely government accounting fictions. (If the government chose to, it could certainly classify some of those tax revenues as recapture of social security, but so what?)

And sure, it's inefficient to cut a check and then have everyone send some back later (although in a world of electronic deposits, it's not that inefficient) there's also the advantage that this form of "means testing" via the income tax should generally work in favor of people who actually get themselves stuck in bad situations- i.e. a rich senior who has a sudden fall in income (perhaps they invested in GM or Chrysler bonds) will automatically be supported through the current situation.

Alternatively, under means testing, the same person would presumably have to go fill out some kind of paper work, have their new and additional benefit levels approved but a complicated and expensive bureaucracy. And how would you measure prospective income anyway- isn't it inherently more difficult than simply seeing what your income actually was over the course of a year? It sure is to me. I guess you could save some money by making the means testing rule based on past income only, but of course that will generate significant hardship exactly the sort of people (those who suddenly lose income) we're setting things up to try and help.

So basically, no offense, this is yet another situation in which our seemingly insane government has evolved to a more rational, flexible outcome than a comprehensive and purely rational approach would likely get to.

Aidan writes:

How can you calculate the cost savings without discussing the additional costs added by implementing means testing? From a Dean Baker study:

"The cost of administering a means test would add substantially to the operating cost of the program. If the means test raised the expense ratio for the retirement program to the same level as the disability program it would increase expenses by an amount equal to 1.70 percent of the program’s cost. This would eliminate most, if not all, of the savings from a plausible means test on affluent beneficiaries."

Hugh writes:

Would SS contributions be reduced as a result of this means-testing?

If not, why not?

To my way of thinking, the fairest pension system is one where you get back what you paid in - plus interest.

If you feel that people at the bottom need more income, this should be paid out of general taxation.

PrometheeFeu writes:

I think there are significant transitional effects. People have planned their retirements based upon Social Security not being means-tested. But set that aside for a moment.

I think the "slippery slope" argument is actually very important. The Social Security tax is a big chunk out of my income. If you take back the promised benefits and keep the tax, I'm not going to be very happy about it. I would imagine, I wouldn't be the only person a little bit ticked off by the idea of having to pay that tax.

Wolfman writes:

I am against any form of "means testing" for SS as long as the current funding method is in place.

Currently, SS is funded through mandatory contributions (confiscation) taken from an employee's pay, justified by the promise of repayments later in life after retirement, or as insurance in the event of incapacitation due to illness or injury.

To suddenly tell someone they aren't going to receive the promised repayment because they've saved too much or have been too successful is just plain wrong. It would amount to nothing more than theft. Embezzlement might be a better word. I would much rather see a gradual increase in the retirement age for full benefits (as has been proposed by Paul Ryan, among others), coupled with the elimination of the cap on earnings subject to FICA withholding.

That would go much further towards making SS solvent than any means testing program. It would also be much easier, politically, to implement.

If you are going to means test SS, which would lead inevitably to denial of any SS benefits once a certain level of income or wealth/assets had been reached, you would need to allow people to "opt out" of the program entirely and give them the option of managing their own retirement. I've read several studies that have shown that simply investing regular amounts in simple index funds and reinvesting all dividends leads to much greater rates of return than you get from the supposed safety of the SS "lockbox", without the need to rely on an increasingly suspect government guarantee.

MikeP writes:

From a Dean Baker study:
"The cost of administering a means test would add substantially to the operating cost of the program. If the means test raised the expense ratio for the retirement program to the same level as the disability program it would increase expenses by an amount equal to 1.70 percent of the program’s cost. This would eliminate most, if not all, of the savings from a plausible means test on affluent beneficiaries."

1.7% reduction in Social Security costs due to means testing, huh? Some means testing!

When I say "means testing", I mean phase out Social Security and Medicare in favor of welfare and Medicaid paid out of the general fund. This is both far more progressive and far less expensive and inefficient.

Cameron Murray writes:

If you are interested in this type of thing in practice, Australia has a form of means testing for our Medicare. Basically, you pay a an extra bit of income tax for income over a fixed amount, 1.5%, if you do not have private health insurance. Otherwise, you pay the little bit extra tax. Health insurance typically costs around $2000 per year.

SheetWise writes:

Here's an idea -- the government allows people to manage their own retirement account, and includes their home as a qualified investment (structuring a mortgage that makes all retirement investment senior to the loan, and requiring reinvestment to another qualified plan on sale).

In a median ~40k household this 12.4% would cover about $5k a year, or over $400 a month. Transferring retirement investment to a home would simply displace current spending, which increases everyone's disposable income. Since the $400 doesn't cover the mortgage for most people -- an increase in SS tax wouldn't even be noticed by the poor, it would just mean more of their home investment was in a qualified plan. It would have the added (dubious) benefit of increasing demand for homes.

This would work. There's just that one pesky problem of how to fund current benefits, since the congress critters can't find any of the trust fund money. We're going to have to bite the bullet, tell the truth, and clean this mess up someday -- why not now?

Larry Willmore writes:

The US internal revenue system allows taxpayers to shelter retirement savings in 401k accounts, adding to the budget deficit. The IRS taxes withdrawals, but only at normal rates. So, why not means-test 401k withdrawals? This would make as much sense as means-testing social security benefits.

Scott Sumner wrote: "Means testing is just another high implicit marginal tax on people who work hard and save."

A 401k thought experiment provides additional evidence for his point. Means tests are equivalent to taxes.

Charlie writes:

It makes me uncomfortable to answer a question about a change in political economy after an exogenous political change. I guess, implicitly, your arguing that status quo bias is strong and if we shook things up, new constituencies would form.

My initial thought is, no they wouldn't. If Bryan got to enact his policy as ruler for a day, the next day new legislation would over turn it.

"In the new political equilibrium, how much do you predict the full Social Security and Medicare benefit will fall?"

So, I guess concisely my answer is, the new legislation is not a political equilibrium, and benefits would rise to the old level, which is an equilibrium.

Ryan Satter writes:

What's up to every , as I am really keen of reading this webpage’s post to be updated regularly. It carries good data.

Comments for this entry have been closed
Return to top