I'm writing a piece for Berkeley's GSPP's magazine, Policy Matters. It's trying to be a positive piece. Basically, I'm advocating Diamond-Orszag plus, where the "plus" includes automatic savings from your tax refund and a stronger Social Security Commission so that the Congress doesn't have to be too brave all at once.
A Genuinely Good Deal for Social Security
J. Bradford DeLong
U.C. Berkeley
March, 2005
Draft 1.2
Suppose that we lived in some bizarre parallel universe in which everything was topsy-turvy, and that as a result we were having an informed debate about what to do with our Social Security system that made sense. I know that this is total fantasy, but bear with me a minute: there is method to my madness. Suppose that this were so: what questions would we be trying to answer?
1. Roughly what proportion of pre-retirement income should we guarantee people will have after they retire--no matter how well their investments do, no matter how thrifty or feckless they were during their working years? In short, how big--measured as a share of pre-retirement income--should the Social Security system be?
We should guarantee a basic Social Security benefit of roughly half of pre-retirement after-tax income--a replacement rate for the median worker, counting state and local as well as federal taxes, of roughly 30 percent of pre-retirement pre-tax income. Anything less runs a substantial risk of producing a lot of elderly poverty: the feckless and the unlucky will, when they are old, live much worse than is common in the surrounding society. Now as society grows richer this elderly poverty will be a relative phenomenon: people will be much poorer than their neighbors, but few of them will be absolutely poor in the sense of living in boxes or eating catfood. Nevertheless, relative elderly poverty is real elderly poverty, and it is something that a good society should protect against.
2. How progressive should the Social Security system be? Those who make less than the average should probably have a higher replacement rate--receive a higher share of pre-retirement income--than those who make more than the average. But how much more?
The Social Security system should be somewhat but not extremely more progressive than is our current system. I see no case for exempting the top 15% of wage income from the Social Security tax base. I see no case for exempting non-wage income from the Social Security tax base. On the benefits side, we already have substantial progressivity: benefits relative to scaled lifetime taxes paid--the "Primary Insurance Amount"--rise at a rate of 0.9 for roughly the first $600 a month in your Social Security check, at a rate of 0.32 for roughly the next $900 a month, and at a rate of 0.15 thereafter. It would be good if the system were more progressive, covered everyone, and offered a minimum benefit to those whose taxes paid in were zero. But otherwise the system seems in good shape as well as progressivity is concerned.
3. How should this basic Social Security system be financed? Should it be pay-as-you-go, in that each generation of taxpayers pays for the last generation's retirement? Or should it be a funded system, that builds up enormous assets and so owns large chunks of the economy, and uses the returns on those assets to finance large chunks of benefits?
The answer to this question depends on the shape of future economic growth and demographic change. When the economy is growing faster than the interest rate, pay-as-you-go systems are attractive: they offer a low-cost way of moving wealth up the generations from the (richer) future to the (poorer) present, and so raising social welfare. When the economy is growing much more slowly than the interest rate, the burdens placed on workers by a pay-as-you-go system are much harder to justify, and funded systems--those that build up lots of assets to finance part or all of this generation's benefits--become much more attractive. Currently, the Social Security Administration's actuaries have a set of assumptions about economic growth, interest rates, and equity returns that I believe are inappropriate, and make pre-funded systems look artificially good and pay-as-you-go systems look artificially bad. There is definitely a strong case for pre-funding some of the cost of Social Security for the large baby-boom generation. There is not a case for prefunding much else of the basic benefit, and in my view there will not be a strong case for prefunding much of the basic benefit until immigration into the United States begins to decline significantly from its current relatively high levels.
4. What additional steps should the government take to make it easy--or perhaps mandatory--for people to save in their own private accounts, so that they reach retirement with more than their basic Social Security benefit to draw on?
This question is, I think, the most interesting. Americans do not take nearly as much advantage of tax-preferred and other savings vehicles as we economists think that they should. I am one of those who believes that America's national savings rate is dangerously low. The bottom half of America's income distribution has essentially no wealth invested in the stock market, and that cannot be right. It seems to me that these are problems that the government should address them.
How should it address them? The government should address them by making add-on savings out of payroll into individuals' private accounts the default--not mandatory, but the default: you have to fill out a form and check out a box in order not to make your contribution to your private account. The government should sweeten the pot: provide a partial match for funds directed into private accounts. The government should also provide a simple and reasonable default option for investing private accounts: half in a low-fee stock index fund, and half in a low-free bond fund. It should offer little else in the way of investment options: the danger that private account holders will be on the least informed side of trades is great, and the danger that private account holders will degrade their account through fees and transaction costs is great as well.
5. What kind of bureaucracy should govern and administer this system? And how much flexibility should it have--what adjustments should it be able to make on its own without going back to Congress for revised authorizing legislation?
It seems clear that the system needs more flexibility than current law allows. Fertility waxes and wanes, economic growth speeds up and slows down, returns increase and decrease. A pay-as-you-go system thought of as a defined-benefit program will always be sliding into deficit or surging into surplus. Americans' entitlement in retirement to their share of pay-as-you-go Social Security revenues is more equity-like than debt-like. Because there is no residual claimant or debtor (besides the U.S. government), the system should be operated more like a credit union or a mutual association, with payouts that naturally rise and fall with resources. Such a system is, I think, best operated with a Social Security Board of Trustees with a fiduciary duty to maintain the long-run actuarial balance of the system, and the power to alter benefit levels (and, within limits, contribution rates) to achieve that long-run actuarial balance.
As for the add-on system, we already have the bureaucracies to run it. The IRS is a natural place to receive the add-on contributions: a check box on form 1040 to opt in--or, better yet, opt out--to the savings program. And an expanded Thrift Savings Plan to manage the money. If it's a good enough system for members of congress and senior administration officials, it should be good enough for all Americans.
Thus the outlines of a potential deal on Social Security--a potential reform--that would be genuinely good for the country are clear:
Shift responsibility for maintaining actuarial balance off of the Congress and onto a Social Security Administration that has added discretion.
Uncap FICA--increase the Social Security tax base to all wage income and perhaps further--and apply the extra resources to sweeten private add-on accounts, to add a little more progressivity to benefits for the poor, and to serve other purposes (like boosting benefits for widows).
Make enrollment in private accounts automatic (it's done automatically on your 1040) but voluntary (you can fill in an extra form to get the money the IRS earmarks for your account back as part of your refund).
Use the government's existing Thrift Savings Plan as a vehicle for managing private add-on accounts--and keep its choices restricted: churning and extra administrative costs caused by asset shuffling are not your friend.
Such a plan should satisfy everyone. It would satisfy optimists who believe SSA's projections are much too pessimistic and that no benefit cuts ever are required: if they are right, it would impose no benefit cuts. This would satisfy pessimists who worry that there is no mechanism to finance the existing level of benefits: if they are right, the SSA will have the fiduciary duty and the power to cut benefits. This would satisfy Congress: if there are benefit cuts, their fingerprints aren't anywhere nearby. This would satisfy believers in boosting national savings: the revenues from uncapping FICA and the money flowing into private accounts from people's choosing the default option will boost national savings. This would satisfy those scared that private accounts would be churned and looted by unscrupulous brokers: the TSP is a good operation that provides powerful protections.
What's not to like?