At first glance, it looks so reasonable. The recent report from the venerable Employee Benefit Research Institute (EBRI) on shifting from the current Social Security system to one embracing what they refer to as individual accounts ("IAs") purports simply to ask "critical questions of how they would work." But what should one make of a study that says that administrative costs under a privatized system would fall somewhere between 10 and 200 basis points? I can easily come up with a private system that costs much more than 200, but it would be meaningless, because no policymaker in his right mind would adopt it.
Not to put too fine a point on it, the EBRI study simply assembles every conceivable objection to privatization, no matter how unlikely, and uses them to engender doubts about fundamental reform.
The timing of the EBRI study is no accident. Over the past year, a broad consensus has developed on key issues in the Social Security reform debate. Everyone now knows, for example, that the current government-run pay-as-you-go system is in serious trouble and cannot pay for promised retirement benefits within the present tax structure. It's also widely acknowledged that most of today's younger workers will get a return of zero or less from Social Security, and trying to solve the problem by raising payroll taxes and/or postponing the retirement age only turns a bad deal into something much worse. Finally, it's abundantly clear that returns in a system of private retirement accounts (PRAs) will be three to five times higher than those promised by the government system.
In short, a system of private retirement accounts may cost slightly more to administer than the current government system. But in return for a modest increase in expenses, such a system would provide a substantial, secure retirement income far higher than that promised by the government system.
One clear measure of this growing consensus is the endorsement of partial privatization by key Congressional Democrats like Daniel Patrick Moynahan and Robert Kerry, and a pretty clear inclination on President Clinton's part to consider the idea.
Skeptics and opponents of a privatized system, recognizing that this consensus has now taken hold, have begun trying to shift the debate instead to the cost (and complexity) of administering a system of PRAs.
The EBRI study is the most detailed example of this approach so far. It offers a laundry list of purported administrative obstacles, including "higher administrative costs" and "significant employer burdens - especially on small businesses." Their bottom line: "The current body of knowledge is too uncertain, and the proposals to date are too vague, to make an objective estimate of how much an IA (individual account) system would cost to administer."
With all due respect to EBRI, our knowledge of the cost of running privately-owned retirement accounts isn't so murky at all. Let's look at what we do know.
The cost of administering the current Social Security system is fairly low (roughly $10 per year per worker), primarily because the IRS system serves as the collection agent, and because there is no money management function.
To say that the government-run system is relatively efficient based on administrative costs alone is similar to saying that the cost of a Trabant (a cheap East German car that never worked) is less expensive that a Taurus or a Mercedes. While true, the fact is that most people prefer to pay more for a car that actually gets them somewhere than to pay less for one that won't. The relative efficiency gains in administering the current system are overwhelmed by the inefficiency of the government-run system in providing the best return for a given contribution.
Administrative expenses in a PRA system will depend on the degree to which Congress and the president impose rules and regulations. If wise decisions are made, administrative costs will be close to or only modestly higher than in the present system. At the same time, for no additional cost, workers will receive the benefit of safe, secure retirement incomes that are substantially greater than under Social Security.
The most efficient PRA system would use existing infrastructure. Over the past two decades, the development and growth of individual IRAs, 401(k)s, 403(b)s, and countless other individual retirement programs has given rise to a substantial infrastructure to handle the sorts of administrative functions also associated with PRAs.
Large firms would find little difference in costs. Their payroll function is automated and there would be few if any costs associated with changing the direction of an individual's payroll deduction from a category that currently gets deposited in the IRS collection system to one where the money is sent to a plan administrator.
For small employers handling their own payroll, the collection problem is only slightly more involved. Instead of sending a check for employee retirement contributions to the IRS each month, the employer could send the check to a certified personal retirement account administrator, along with a form showing the wages and retirement contributions for each individual worker. Since each employer already calculates this information in order to determine the total Social Security contribution, there is only a relatively small additional burden placed on the small employer.
To keep costs at a reasonable level, certified administrators could be required to provide their services to all businesses and charge the same basis point cost for all individual accounts, in which case the relatively low cost of administering large business accounts would subsidize the accounts of smaller businesses.
Money management fees might be reduced substantially if the new PRA regulatory body were to limit investments to passively managed index funds. Administrative and money management costs can easily be pro-rated on the basis of overall assets. This would avoid the situation whereby a new smaller account might have the entire first year's return wiped out by a fixed administration fee.
Based on existing plans and on discussions with industry representatives, I believe it is likely that the range of administrative and money management expenses for PRAs could reasonably be expected to fall between 30 and 60 basis points of the assets under management. At the low end of the expense range (30 basis points), total costs would be close to $10 per worker, similar to Social Security. At the high end of the range (60 basis points), the costs of PRAs would be close to $20 per worker.
The lower end of the range is more likely because of two factors: the large size of a system of PRAs, creating significant economies of scale; and a presumed limitation of money management to passively managed accounts.
In short, a system of private retirement accounts may cost slightly more to administer than the current government system. But in return for a modest increase in expenses, such a system would provide a substantial, secure retirement income far higher than that promised by the government system. And what employee wouldn't jump at the chance to buy a Mercedes for the cost of a Trabant? As awareness of the appeal of a privatized approach to retirement grows in the months to come, EBRI and others resisting the idea would be well advised to stop throwing up roadblocks, and start using their considerable experience and expertise to make the private account alternative work.