Bryan Caplan  

Means-Testing Social Security: The Cohen-Friedman Debate, II

PRINT
Why Minimum Wages Can Take Tim... Henderson on the Case Against ...
As you may recall, Milton Friedman debated former HEW Secretary Wilbur Cohen on means-testing Social Security back in 1972.  Now let's check out the Friedman highlights.  The Great One begins with his characteristically brilliant contrarianism: Public opinion notwithstanding, Social Security, not welfare, is the policy disaster.
Widespread dissatisfaction with the public assistance program-with the so-called welfare mess-has produced numerous proposals for drastic reform, including the President's proposed Family Assistance Plan now before the Congress. On the other hand, general complacency about social security is reflected in pressure to expand it still farther.

My own attitude toward the two programs is almost the reverse. Bad as the welfare mess is, at least public assistance does go mainly to needy persons who are at lower income levels than the persons paying the taxes to finance the payments. The system badly needs reform but, at the moment, it serves an essential social function. It seems impossible to eliminate it promptly, even though its elimination should be our long-term objective. On the other hand, social security combines a highly regressive tax with largely indiscriminate benefits and, in overall effect, probably redistributes income from lower to higher income persons. I believe that it serves no essential social function. Existing commitments make it impossible to eliminate it overnight, but it should be unwound and terminated as soon as possible.
Friedman's probably wrong about the net redistributive direction of Social Security, but his core point is sound: Taxing everyone to help everyone is madness.  More:
I believe that a program which is going to give income to people, which is going to give funds to people, should have a means test. I believe we have a responsibility to the taxpayer and not only to poor people.

I believe that the person who pays taxes has every right to require that, if he pays the taxes in order to help somebody, there be some evidence that that person needs help.
Friedman's ultimate response to Cohen's "A program for the poor will most likely be a poor program" argument:
A program of that kind would be vastly less expensive than the ones we've now got. It would do a better job of helping people. In my view, the task of people like Mr. Cohen and myself is not to speculate about what people will do if they don't have leadership but to try to provide leadership in order to obtain the kind of good program that would achieve our objectives.
Overall verdict: While I'm glad Friedman takes my side on means-testing, his arguments are disappointing.  He doesn't clearly compare - much less measure - the disincentives of means-testing versus the disincentives of taxes high enough to fund universal programs.  He doesn't ponder alternative measures of "means" (income, wealth, expected lifetime income, etc.).  He doesn't emphasize the obvious fact that most Americans are perfectly capable of saving for their own retirement.  And he doesn't drive home the overlooked fact that retired home owners with zero cash income only look poor.  They can and should be expected to support themselves out of their home equity.

P.S. If you think of Friedman as a strict Chicago-School utilitarian, think again:
Mr. Cohen's view, one which unfortunately is very widely held today, is different. It treats the nation not as a collection of individuals and of the groups which individuals separately value, but as an organic unit. In his paper, Mr. Cohen says: ttl speak for the view that the nation should be generous toward the elderly." I would argue that the nation can't be generous to anyone. Only people can be generous. Generosity is a human, individual trait, not a collective trait. There is no generosity involved in my imposing taxes on you to help him. That is not generosity.




COMMENTS (18 to date)
jon writes:
He doesn't emphasize the obvious fact that most Americans are perfectly capable of saving for their own retirement.
The even more obvious fact, and the reason we do have Social Security -- taxing everyone to help everyone is NOT actually madness, sorry -- is that most Americans won't save for their retirement, despite being capable of doing so.
foosion writes:

Unless you cut Social Security benefits for those with relatively modest incomes, you wouldn't save enough money to really matter.

Without the support of the middle class, Social Security wouldn't survive. Meaningful means testing (saving enough money to really matter) would result in big cuts for the middle class and therefore loss of support for the entire program. If that's the goal, at least be clear about it.

the disincentives of taxes high enough to fund universal programs

It's far from clear that those disincentives are high enough to matter for the economy or that there are negative net effects from taxing the best off to fund the less well off.

ThomasH writes:

I understand the objection to financing SS and Medicare from a wage tax because of the disincentive to work/employ, but if financed from a progressive consumption tax (that would also diminish the disincentive to save) what would be the objection ("madness") to SS Medicare spending?

Greg G writes:

Social Security was never intended to provide 100% of people's retirement needs. It was intended to try and reduce poverty among the elderly and it has been very successful in doing so. In fact it is one of the most successful and popular programs in the history of government. But, if you hate government enough, you will hate its successes even more than its failures.

Social Security is one of vanishingly few government programs which was actually passed with a tax that funded the spending approved in the legislation. Compared to other government programs, it has been a model of fiscal responsibility.

Social Security is somewhat redistributive. Those with lower incomes do tend to get larger benefits in relation to what they paid than those with higher incomes.

One reason the program was made available to all workers, rather than only those in poverty, is in order to avoid the perverse incentive of having it only reward those who failed to save for their own retirements.

James writes:

Greg G,

"One reason the program was made available to all workers, rather than only those in poverty, is in order to avoid the perverse incentive of having it only reward those who failed to save for their own retirements. "

Is that a sufficiently good reason in your mind? Restricting SS to those who become poor in their old age might discourage people from saving for retirement but this has to be weighed against the incentive effects caused by the high taxes needed to make SS universal.

" In fact it is one of the most successful and popular programs in the history of government."

You say it's popular. What percentage of people do you belive would participate if it were voluntary? On that note, you clearly like social security and if it were up to me I would never propose depriving you the choice to participate in it. If it were up to you, would you give me the freedom to opt out?

Greg G writes:

James,

Of course it's not really up to me, but if it was, I would not be in favor of allowing people to opt out of social Security. Nor would I be in favor of allowing them to opt out of paying for a share of national military defense, policing and the justice system. In all these cases, I believe that the people who would want to opt out would end up mostly free riding on the share of those costs paid by others.

If you opted out and wound up in poverty as a result, you would likely wind up back on some form of welfare.

matt H writes:

"You say it's popular. What percentage of people do you belive would participate if it were voluntary? On that note, you clearly like social security and if it were up to me I would never propose depriving you the choice to participate in it. If it were up to you, would you give me the freedom to opt out?"

Probably not. If it were voluntary only the very richest would opt out. Maybe the top 3%.

Additionally, if the market crashed as you retired and you were too infirm to work, I suspect you might want government assistance. What about if in your old age you just planned badly and lived too long?

All evidence says the public would not let you starve. The truth is we are going to have some sort of welfare state, because both bad luck, imprudence, and empathy all exist and are common.

Maniel writes:

Prof. Caplan,
That SS exists at all is a tribute to two concepts: retirement worship; and old people are poor. When government makes first claim on my earnings (the payroll tax), they lock away my “discretionary money” (in the sense that if I really needed it, I would not be able to survive without it). Unfortunately, discretionary money is what most of us need to put a down payment on a house, invest in our health, help our children pay college expenses, and save enough to even consider retirement. Most of us use our discretionary money wisely enough so that we will not be poor when older and will be able to afford retirement; for the others, I agree with Prof. Friedman
The impact of the loss of discretionary money is visible in lower levels of savings and home ownership and increased levels of debt, both public and private. And as we all know, demographics will have the last word on this issue.

KevinDC writes:

Greg G,

You say "One reason the program was made available to all workers, rather than only those in poverty, is in order to avoid the perverse incentive of having it only reward those who failed to save for their own retirements."

It seems to me that this argument proves too much. Applied consistently, wouldn't it be equally valid to eliminate means testing for literally any government program - "One reason to make food stamps and welfare available to all workers, rather than only those in poverty, is in order to avoid the perverse incentive of having it only reward those who failed to hold a steady job." Do endorse that view? And if not, what, specifically, is the principle that allows you to accept the claim you made but reject it when applied to other welfare programs?

Further, you say "I would not be in favor of allowing people to opt out of social Security. Nor would I be in favor of allowing them to opt out of paying for a share of national military defense, policing and the justice system. In all these cases, I believe that the people who would want to opt out would end up mostly free riding on the share of those costs paid by others."

This comparison is faulty. Free rider problems occur when benefits are non-excludable. If I stopped paying toward the military, the U.S. military still prevents the Canadians from invading my apartment. If I stop paying towards the police, there are still police in my neighborhood and I benefit from the deterrent effect of their presence whether or not I pay. But social security benefits just aren't like military or police protection. They're easily excludable - hence there's no comparison here.

I expect your response would be to say that sure, we *can* exclude people from collecting social security payments (or other welfare payments in lieu of social security), but we *won't*. But if so, this is textbook question begging. You're simply stipulating that benefits must be made non-excludable and that nobody should be allowed to opt out, and then saying that compulsion in preventing people from opting out is justified because people will free ride on the non-excludable nature of these benefits. You've built the "justification" into the premise. You can't design a system perfectly built to create a free rider problem and then declare that same system to be justified by the free rider problem it creates.

Matt H asks "Additionally, if the market crashed as you retired and you were too infirm to work, I suspect you might want government assistance. What about if in your old age you just planned badly and lived too long?"

Well I'm not anywhere near retirement age (my parents just crossed that threshold), but in my youth I've found myself in personal financial crisis many times, due to various combinations of stupid choices and plain old bad luck. But it never seemed to me that this meant other people were obligated to pick up my tab. People helping each other out is a great thing - I've benefited from the generosity of friends and family and I have taken lengths to help those around me in need when I've had the ability to do so. This is a beautiful thing. But I just can't get from here to the idea that "It's my right to receive financial benefit from other people because I had bad luck and/or made bad choices." There's a gigantic chasm between those two concepts. How exactly do you bridge that?

Greg G writes:

Kevin

I don't think that general principles always apply equally to specific issues - especially in cases where motivations and incentives are very complex. There are always trade offs. The EITC has been a quite successful attempt to address the issue you raise about welfare benefits reducing the value of holding a job.

I think there are some important non-excludable benefits of keeping more people out of poverty. I think neighborhoods are safer and more pleasant places when more people are kept out of poverty. I think there are more business opportunities for entrepreneurs when a larger percentage of the population are consumers. I think that you will benefit if fewer of your friends and relatives need to come to you for handouts if they fall into poverty.

I am glad that you have had the good fortune to have friends and relatives who have been able to help you out when you have fallen on hard times. Not everyone is so lucky.

Social Security requires most people to pay most of the cost of keeping themselves out of poverty in their old age. It is actually less redistributive than most other Federal programs. Means testing it would make it more redistributive. If you are opposed to redistribution you should be opposed to means testing.

KevinDC writes:

Greg G,

I agree completely that "general principles [don't] always apply equally to specific issues" - that's why we call them *general* principles :-) Still, your response on that point seems inadequate to me. When general principles seem to not apply, it's important to spell out and understand what are the morally relevant factors which create the distinction. If we abdicate the responsibility to explain *why* a given case is special, then nothing prevents us from just hand-waving away any and all principles by just saying "General principles are imperfect and not universally applicable - no further case is required." Exceptions to the rule certainly exist, but when a person is claiming an exception the burden is on him to make that case.

"I think there are some important non-excludable benefits of keeping more people out of poverty." I can think of all kinds of programs which would create nonexcludable benefits. A law which required people to do regular yard work, for example. Neighborhoods would look nicer, which leads to higher property values for all homeowners (the nicer your neighbor's house looks, the better your property values - and he can't exclude you from that), and (according to the broken windows theory, anyway) it would likely cause a drop in the crime rate to boot. All of these benefits are nonexcludable. But I'd still oppose a law forcing people to do yard work. Just like you can always find externalities for any action if you look hard enough, you can find nonexludable benefits from any program if you look hard enough. It just doesn't follow that if a program creates benefits which are nonexcludable, even substantial ones, that the program is therefore justified.

"I think that you will benefit if fewer of your friends and relatives need to come to you for handouts if they fall into poverty." Well...that's a bit of a bloodless attitude. I don't view helping out my loved ones in their time of need as some kind of burden I need to offload onto taxpayers. But if we are to take an accountant's attitude about it...it would be a financial benefit (ugh) to me if I didn't have to give "handouts" to people I care about who are in need, but it's also a financial burden on me to be taxed out of every paycheck to provide programs to complete strangers. So...I'm not sure what your point here is? Maybe you're assuming the amount I'll be taxed is less in the long run? Even if true, I don't see why this should provide me with a reason to support such measures. "Policy X would be financially beneficial to me" =/= "Policy X is justified."

"I am glad that you have had the good fortune to have friends and relatives who have been able to help you out when you have fallen on hard times. Not everyone is so lucky." True, and Paris Hilton was luckier than me by being born into an incredibly wealthy family. That doesn't give me some sort of claim against her. (Unless you find Rawls original position persuasive?)

"If you are opposed to redistribution you should be opposed to means testing." I'm not totally opposed to redistribution in principle, though I often oppose it in practice. But Social Security is a particularly strange system. It's true that the more you pay into it, the more you collect at the end. Perhaps you view this as a feature. To me it's a bug that borders on insanity. It means that the people who had the highest lifetime earnings, and were therefore in the best position to provide for their own retirements, also receive the highest amounts of taxpayer "assistance" when they reach retirement age. That's just batty. Warren Buffett will collect larger social security checks than my lower middle class parents. Sure, he paid more into Social Security than they did, but he also paid more into food stamps. Does that mean we should give rich people food stamps, because doing so makes food stamps less redistributive?

(Also, is "redistributive" a real word? It's a funny sounding one if it is.)

Greg G writes:

Kevin,

I take your point about the need to make our cases when claiming an exception to principle. It's a fair criticism.

Even so, I expect we differ a lot in how much we think general principles decide these issues. The devil is always in the details. It's always a judgment call. Lots of general principles apply to each case and they often conflict.

I agree that non-excludable benefits do not automatically justify legislation. And that they are easy to find. And that we should not legislate in favor of compulsory yard work. (We are really building bridges here.)

I only brought up non-excludable benefits for Social Security because you suggested they were impossible to find.

I am glad to hear that you have never found extending needed help to friends and relatives to be a burden. It can become a burden. I have been a Hospice Volunteer for 18 years and I have seen it become a burden to family members many times. I understand that you meant you would accept that burden without complaining or shirking and I believe you. Most people do. Seeing that is one of the best parts about being a volunteer. It is inspirational. But sometimes one family member outlives all the others after being bankrupted by the illnesses of those who died earlier.

I'm sure I like Rawls a lot more than you do but, in the end, I think he relies too much on too few general principles too.

As for the legitimacy of the word "redistributive" the internet does recognize it but that's a fairly low bar. I happily admit to not having heard it before I wrote it but I maintain it did everything I would want a word to do.

James writes:

Greg G,

I asked you about opting out of SS but you changed the subject to opting out and later going on welfare. Also, I notice you still didn't address the matter of comparing the incentive effects of means testing versus the incentive effects of having high enough taxes to make SS universal. Is there a reason for these lapses?

I wish you would see the real free rider problems in social security as clearly as you imagine the free rider problems that come with letting people opt out. Right now, you value SS and I don't but we both have to participate. Guess who's free riding on who?

Matt H,

Your estimate of 3% seems a bit off. Where SS isn't available, people could put 12% of their income into savings vehicles with rates of return close to zero. Fewer than 97% do.

Greg G writes:

James,

You asked me if I favored making Social Security optional. I said that I didn't and I gave you the reasons why. I think of that as a disagreement, not a "lapse."

The reason I didn't compare the different incentive effects you ask about is that I don't think there is any reliable way to measure them. If you think you have some reliable way to measure these effects I would be very interested to hear about it. It comes down to a disagreement about how some counterfactual that we are never going to see might play out. Apparently I am more OK than you are with us disagreeing about that.

I get that you value the benefits Social Security provides less than most people. There are no government programs that are equally valued by all people or supported by all people. Do you think pacifists should be able to opt out of paying for national defense? Should anarcho-capitalists be able to opt out of paying for the justice system?

KevinDC writes:

Greg G,

"I only brought up non-excludable benefits for Social Security because you suggested they were impossible to find."

Well, whether they are impossible to find depends on how we are defining our terms, which is what unfolded in our conversation so far. As a parallel, think of the standard economic model that says we should tax negative externalities and subsidize positive externalities. Pretty much any activity will have both, as long as you look hard enough and define them broadly enough. So, this leaves us with two options. We can define "externalities" in vastly broad terms and come to the conclusion that we should be taxing and subsidizing everything simultaneously. Or we can take a narrower definition of externalities that limits it to clear cut and unambiguous cases.

If we try to justify Social Security on a free rider problem, based on the nonexculdable nature of these benefits, the same options exist. If we're defining nonexcludability in a clear and straightforward way, this justification fails outright - sending someone Social Security benefits is easily excludable. It is, of course, "possible" as you say to find nonexcludable benefits, but only by defining such benefits in a way so broad, vague, and hopelessly plastic that this definition can be made to fit anything under the sun.

"I expect we differ a lot in how much we think general principles decide these issues." Probably less than you think.

"The devil is always in the details. It's always a judgment call. Lots of general principles apply to each case and they often conflict." I 100% agree here. This itself is a principle I adhere to quite strongly, which is one thing that keeps me inoculated from being a Rothbardian. (Not the only thing, mind you. Just one thing.) Call it the Principle of Conflicting Principles. Still, like all principles, it can be subject to misuse. Invoking the Principle of Conflicting Principles, properly used, should be a charge to think really hard about whether or not our case is justified. But it's all to tempting to use it instead as a trump card against doing the work needed to justify our case. In other words, the PCP should be a starting point, not an end point.

It seems to me you've been using it that second way in our discussion so far. You said giving people the freedom to opt out of Social Security creates a free rider problem. I said not so, because Social Security benefits are easily excludable and thus not subject to the free rider problem. You responded by giving a very broad and vague list of nonexcludable benefits you think Social Security could provide. I in turn pointed out that such a broad definition of nonexcludable benefits could easily be used to justify pretty much anything you can think of, so if they justify Social Security then why not these other programs? Invoking the PCP here doesn't do the the legwork needed to answer this question - it's an acknowledgement of the work still ahead of you, not an abdication of the responsibility for making your case. Perhaps I'm being uncharitable, but it seems like that's what you've been doing. I'm asking "What's the morally relevant difference here? Why does it justify A but not equally justify B?" And it seems like you're saying "Well, the world is really complicated and full of imperfections and trade offs and conflicting values," and just leaving it there. That statement is true, but just leaving it at that is not an answer to the question. It's an excuse to dodge the question.

You agree that (broadly defined) nonexcludable benefits are easy to find, and that they don't intrinsically justify legislation. I maintain that you can't at this juncture simply shrug your shoulders and make vague claims about the complexity of the world. True, grey areas exist. Grey areas will always exist. It doesn't follow from this that therefore everything is a free for all. The more complex and complicated the world, the more carefully we must examine and justify our cases. This is especially true when coercion is involved.

Anyway, I need to get some actual work done at work. They don't pay me to look pretty! (And thank the gods for that too, because I have had plenty of being poor for a lifetime already.)

Greg G writes:

Kevin,

I made the argument. You just ignored it. That's why I pointed out that taking care of dying relatives can, and often does bankrupt people who do the right thing and put everything they have into taking care of their families. Such cases would not distinguish between who has opted in and out of Social Security.

I think we have a collective as well as an individual responsibility to see that people in desperate need are taken care of. And to consider policies that are efficient and effective at keeping people out of need in the first place. We are social animals. I think basic principles of fairness are violated when those in great need aren't taken care of. That is another principle that sometimes conflicts with the desire to avoid coercion.

I am well aware that most commenters here disagree with me and feel they occupy the high moral ground. I concede their taxes could be lower if there was no taxpayer funded social safety net. I know it's impossible to legislate more compassion. It is sometimes possible to legislate more fairness. I think Social Security does that. I am aware that is not a popular position on this blog.

Floccina writes:

IMHO the best thing to do with SS is give everyone over 67 years old $200/week. That would cut the program with minimal change in incentive to work and save.

Robbl writes:

Bryan,
You say that a retired person's equity will finance their retirement with no cash income. Can you point me to an article that shows how to to do that? I would sleep a lot better at night.

Right now I have lots of cash and equity, but lacking relatives to care for me when when dementia sets in the only logical place to live is a continuing care facility which requires a very large income...

Comments for this entry have been closed
Return to top