Bryan Caplan  

Universal Social Programs vs. Cost-Benefit Analysis

PRINT
Krugman on Ponzi: I Was Being ... Willpower...
Economists habitually mock protectionism for its high cost-benefit ratio.  "$265,000 per job saved!  How ridiculous."  But when you propose means-testing Social Security and Medicare benefits, these same economists usually protest, "A program for the poor will always be a poor program.  The only way to make well-funded programs popular is to make them universal." 

Notice, however, that we can easily ridicule universal social programs in exactly the same way that we ridicule protectionism.  Suppose two-thirds of the population is perfectly able to provide for its own retirement and health insurance.  Then the budgetary cost of cloaking welfare for the bottom one-third in universal garb is triple the apparent cost.  And that ignores all the disincentive effects of the extra taxes and giveaways.  Add it all up, and you could easily get a number in the protectionist ballpark.  Think: "$100,000 per retiree lifted out of poverty!  How ridiculous."

The defenders of universal programs could protest that, "Political forces tie our hands; there's no other way to help the elderly poor."  But couldn't the protectionist plausibly make an analogous argument?  American voters don't like the idea of the Treasury mailing $200k checks to displaced auto workers; tariffs and quotas to revive the auto industry just feel better.

I suspect that if push came to shove, most economists would respond, "Since there's no cost-effective way to help displaced auto workers, the least bad solution is not to help them."  So why are they so reluctant to say the same about Social Security and Medicare?  If cost-effective means-tested programs are politically unsustainable, isn't it possible that the least bad solution is no programs at all?


Comments and Sharing





COMMENTS (9 to date)
Lorenzo from Oz writes:

Speaking as an Australian, this debate seems very strange. We means-test almost all income transfers, making them fiscally more sustainable and it seems to make them easier political sells.

Brian writes:

The cost of Social Security is well under 1% of the money transferred. The SSA is extremely efficient partly because it doesn't need to audit incomes, analyze tax returns, uncover hidden assets, spy on recipients, or do any of the other obligatory means-testing bureaucracy.

Sure, three times as much income is transferred as would be transferred by a program that only addressed the poorest third. But two-thirds of that income is transferred right back into the pockets of the richest two-thirds so their net average lifetime tax paid to Social Security is zero. Over a lifetime, Social Security tax costs the richer two-thirds much less than a means-tested program would cost because they don't have to pay for means testing.

It's possible the richer fraction would like some of their retirement savings in a riskier potentially-higher return vehicle. They can still put it there because they are rich enough to save above and beyond Social Security. Most of them would like at least some really low-risk fraction and they get it automatically.

So really, Social Security costs the rich almost nothing. It saves them the cost of running a means-tested program. It gives them flexibility to invest in riskier modes. It's best for both the rich and the poor. It is suboptimal, however, for the high-paid army of government bureaucrats that would run a means-tested program.

Social Security is utterly unlike protectionism. Protectionism costs everyone in the society a giant deadweight loss. It provides relatively small benefits to a visible few. Social Security has approximately zero net cost to the rich because they get back their payments. The poor benefit from income security. The deadweight loss of Social Security roughly zero, much less than that of a means tested program or protectionism.

Frank Howland writes:

The standard line from economists, which in this case is correct, is that the budgetary costs are not the costs to pay attention to, only the disincentive costs--higher tax rates resulting from the need to raise revenue are probably most important, though the program may introduce further inefficiencies (e.g., Social Security and saving)--are relevant. The bulk of the budgetary cost is a transfer from one group of people to another, but doesn't involve overall economic welfare falling.

Protectionism is indeed different from social insurance, as Brian points out in the post at 2:32 AM. A much higher deadweight loss to money spent ratio.

ed writes:

Frank is of course correct. As an economist, you should be concerned with economic cost, not budgetary cost.

I'm surprised to see Bryan make this mistake.

The economic costs of SS are hard to measure. They include any deadweight loss of taxation (which in theory should be less than for other taxes because of the link between taxes paid and benefits received.) Also there is a possible loss in growth due to lower investment, because of lower incentives to save (Martin Feldstein's critique).

Dan Hill writes:

@Brian said "two-thirds of that income is transferred right back into the pockets of the richest two-thirds so their net average lifetime tax paid to Social Security is zero."

Well only if you ignore the fact that the government has been taking money from them their whole working life, preventing them from investing it themselves, not investing it in any meaningful sense of the word and creating not even a legally valid asset, but only a politician's promise that the money will be eventually returned.

TomO writes:

@Dan Hill said "Well only if you ignore the fact that the government has been taking money from them their whole working life, preventing them from investing it themselves, not investing it in any meaningful sense of the word and creating not even a legally valid asset, but only a politician's promise that the money will be eventually returned."

But he is not ignoring the investment themselves cost. He is saying that you need to discount that cost against their need to have a more conservative investment portfolio in a world in which they don't receive SS benefits.
So while its not as simple as SS cost = (SS taxes paid - SS benefits paid - SS tax as transfer to poor people). It isn't SS cost = (SS taxes paid - SS tax as transfer) either.
Instead you would calculate the difference between SS benefits paid vs. the expected avg. return on SS contributions (where that = SS taxes paid - the portion for transfer). And you would need to lower your expected avg. return as avg. person would need to invest more conservatively to compensate for lack of guaranteed SS benefits.
It is argued that the difference amount is smaller than the costs of administering a means test.
Even if it isn't the best solution isn't to means test - but to cut SS returns for high earners with a corresponding across the board cut in SS taxes.
(Sadly neither that nor means testing are likely politically feasible)

Floccina writes:

If ask people why Social security exists they will tell you that it exist so that no one will be destitute in retirement.

Here are some excerpts from the Social Security web site:

The maximum benefit depends on the age a worker chooses to retire.  For example, for a worker retiring at age 66 in 2011, the amount is $2,366.  This figure is based on earnings at the maximum taxable amount for every year after age 21.

The average monthly Social Security benefit for a retired worker was about $1,177 at the beginning of 2011.  This amount changes monthly based upon the total amount of all benefits paid and the total number of people receiving benefits.

BTW I leaned some thing from doing this research that is post.  I learned that there is no minimum benefit from Social Security so it does not even cover some of the neediest people.

So if we want to keep people from becoming destitute in retirement but want to be more efficient and not discourage work and saving what can be done with Social Security to save money?  You could means test the benefit in retirement but that discourages saving.  What you could do is select an amount below the average benefit and give that to everyone.  So how about we give everyone on the program $800/month that would reduce spending on the program by of over 32 percent.

[disallowed formatting removed. --Econlib Ed.]

Noah Yetter writes:

Never heard an economist not named "Paul Krugman" say "A program for the poor will always be a poor program." Politicians and journalists say that, not economists.

Philo writes:

It might be replied that *no program at all* is itself politically unsustainable. But I would not so reply, because I do not understand what "politically sustainable" means. Whatever program (or non-program) is actually adopted should obviously count as "politically sustainable," but I presume the concept is supposed to extend beyond the actual. Some programs that are not adopted are nevertheless supposed to be PS, though others are not. But where's the dividing line?

Comments for this entry have been closed
Return to top