Monday, March 31, 2008

The cost of "sitting on" Social Security

In a Seattle Times column by Froma Harrop, Dean Baker is quoted regarding his preferred plan for Social Security: wait and see how big a problem it turns out to be.

What should we do about Social Security? "I would just say, 'Let's sit on this,' " Baker answers. If come 2030 Americans see problems looming, he adds, "we can do something."
Here is the Social Security Trustees best guess of what "sitting on it" until 2030 would mean. The current 75-year Social Security deficit is 1.7% of payroll, meaning that and immediate and permanent increase of 1.7% in the payroll tax, from 12.4% to 14.1%, would be sufficient to restore solvency for 75 years. (An equivalent immediate and permanent benefit reduction would do the same.)

With each passing year, an additional year of deficits is added to the 75 year calculation, thereby increasing the size of the deficit. This tends to increase the shortfall by around 0.06% of payroll each year. Therefore, our best guess of the 75-year actuarial deficit in 2030 would be around 1.32% of payroll higher than today's, for a total of around 3% of payroll. This would require an immediate tax increase of 3%, to 15.4%, or equivalent reduction in benefits.

Moreover, while there is a great deal of uncertainty regarding Social Security's future finances, this uncertainty isn't a reason for delaying action. Remember that things could turn out worse than the Trustees project, not only better than projected, and people would be willing to act sooner or even over-balance the program in order to avoid this unlikely but adverse outcome.

Read more!

Sunday, March 30, 2008

Paul Krugman on trust fund, with comments from Andrew Samwick

Paul Krugman comments on the latest Trustees Report while Andrew Samwick, from his new blogging location, puts Krugman's claims to the test.

Krugman's basic argument -- which I believe really originated with Dean Baker -- is that however we characterize the trust fund, calling Social Security a "crisis" is incorrect. If we believe there is a trust fund, Krugman says, then the program is solvent until 2041, making it a significant but not-so-pressing problem. If we don't believe there is a trust fund, then Social Security is just a part of the federal government and so is funded in perpetuity.

Clever, but limited for several reasons. For once, consider that much of Medicare is not financed through a trust fund, being paid entirely through general revenues, but that doesn't mean we don't consider its cost growth a problem. (It's the only problem, some would argue.) Likewise, Social Security's costs are projected to rise from 4.3% of GDP today to 6% of GDP in 2030. Those costs must be paid, whether we consider trust fund real or not. Whether we consider the trust fund real may influence how we think they should be paid, but the cost increase is real and significant.

Samwick takes on this quote from Krugman:

"As Kevin Drum, Brad DeLong, and others have pointed out, the SSA estimates are very conservative, and quite moderate projections of economic growth push the exhaustion date into the indefinite future."
Samwick points out, much as I have (see here and here), that this claim isn't particularly plausible:
You can look at the sensitivity analysis for the growth in real wages in Table VI.D4 and see that increasing the projected rate of growth in real wages by 0.5 percentage points (around the baseline growth rate of 1.1% per year) shrinks the 75-year actuarial deficit from 1.70% to 1.12% of taxable payroll. That gets us about a third of the way toward a zero balance over 75 years and is a necessary but not sufficient condition to support Krugman's claim. If continued linear extrapolation is valid, then we would need to add about 1.5 percentage points to the real wage growth rate--over 75 years--to get the balance to zero. That's sustained real wage growth of 2.6% per year for 75 years. Krugman should come out and say that such a number is "quite moderate" if that's what he means. Seems pretty optimistic to me.
For a fairly lengthy alternate take on what it means for the trust fund to be "real", see here. Read more!

Tuesday, March 25, 2008

2008 Trustees Report released

Today the Social Security Trustees released the 2008 Report on the program's finances. The report is available here. The summary of the Social Security and Medicare Trustees Reports is available here.

The short story is that the short- and medium-term forecast is about the same, while the long (and extra long) term forecast is improved. The date at which social security begins to run cash deficits remains the same at 2017, as does the date when the trust fund is projected to become insolvent (2041).

However, the 75-year actuarial deficit improves significantly, from 1.95 percent of taxable payroll to only 1.7 percent of payroll. One way to think about this is that an immediate and permanent payroll tax increase of 1.7 percentage points -- from 12.4 percent to 14.1 percent -- would be sufficient to keep the program solvent for 75 years, though not beyond.

Beyond 75 years, the infinite horizon actuarial deficit also declined, from 3.5 percent of payroll in the 2007 Report to only 3.2 percent of payroll in the current Report.

These long-term deficit reductions reflect not changes in the Trustees' assumptions, but improvements in the methods the SSA actuaries use to project the program's finances. Specifically, the actuaries improved their modeling of immigration, particularly of individuals who immigrate to the United States but later emigrate, generally to their country of origin. These individuals would work and pay taxes into Social Security, but generally not collect retirement benefits. As a result, they are a net plus to system financing.

Commentary: While these new methods for modeling immigration clearly correct a shortcoming in previous methods, more remains to be done in modeling the earnings, fertility and mortality of immigrants, who often differ in all these respects from the native born. In general, the earnings, fertility and mortality of immigrants are not modeled distinctly from the native born. Lower earnings and lower mortality means that immigrants generally receive a higher return from Social Security. Lower earnings also means that individuals who work in the U.S. but return home are a smaller windfall, since their taxes would be lower than average American workers. However, higher fertility could mean that future population growth would increase, which would improve Social Security's finances. I believe these issues can best be analyzed using a microsimulation model in which individuals are modeled distinctly, rather than in the semi-aggregated cell-based model the SSA actuaries use for most of their analysis. Read more!

Monday, March 24, 2008

Life expectancy gap between rich and poor widens

The New York Times reports that while average life expectancies are increasing, they are rising at different rates for Americans of different incomes. The gap in life expectancies between rich and poor is growing. The Times story highlights a number of potential causes, ranging from different risk factors, different levels of medical treatment, and differences in long-range planning for individual health. All likely play a role.

A factor the Times article does not mention is that causation between income and health runs both ways. While those with lower incomes can’t afford as good health care, it’s also the case that those with poorer health tend to earn less, simply because they cannot work as much. If poor health has a stronger negative effect on income today than in the past, this could help explain the trend.

Differential mortality by income has been an issue for Social Security policy for many years. Milton Friedman argued decades ago that Social Security’s formal progressivity was reduced due to the different life spans of the rich and poor. (This is true, though the program remains progressive overall.)

More recently, Peter Orszag and Peter Diamond argued that the Social Security benefit formula should be adjusted to account for rising differential mortality by income. Their reform plan indexed all Social Security benefits to account for rising average life expectancies, but also increased the progressivity of the benefit formula to account that longevity was rising faster for the rich than for the poor.

Read more!

Sunday, March 23, 2008

Social Security Trustees Report Tuesday

This coming Tuesday the Social Security Trustees will release their annual Report on the financial status of the program. When the Report is released it should be available here.

During the almost five years I spent at the Social Security Administration I was involved with the preparation of the Trustees Report. By the time I became Deputy Commissioner I led the SSA staff who took part in the Trustees working group, which is the day-to-day staff level group that effectively prepares the Report. As I left the agency only in February I was involved in the preparation of the 2008 Report, but like current staff will not discuss the results of the Report prior to its release.

However, this might be a good opportunity to briefly discuss the Trustees process, which I never understood particularly well before I became involved with it. Several issues may be of interest.

First, the White House is almost totally insulated from the process of producing the Trustees Report. White House staff are not involved with the Trustees working group in any way and do not attend Trustees meetings. The White House is not told of changes in the Trustees assumptions and does not know the overall results until the Report itself is released. This is designed to limit political influence on the Trustees Report.

Second, there are often questions of whether the Trustees themselves attempt to exaggerate the size of the problem, either to accomplish political goals or simply to push the public toward action. In the time I spent working on the Trustees Report I never witnessed anything at all along those lines. In fact, it is refreshing how professionally the staff, both political appointees and career, go about their business. I have on several occasions witnessed political appointees arguing for changes in assumptions, purely on the merits, that would make the actuarial deficit smaller. Likewise, I have seen career staff arguing for changes in assumptions that would tend to make the measured deficit larger. The are obviously disagreements about both assumptions and methods, and how these disagreements are resolved would tend to affect the size the of deficit. But among the individuals involved, I have found that these disagreements tend to be normally distributed – that is to say, the fact that a given person holds a certain view that would make the measured deficit larger does not mean that all of his/her views would do so. In short, in my experience issues tend to be examined on the merits.

Third, many people don't understand how strong is the role of the SSA actuaries (who are career staff) within the Trustees Report process. Technically, the Trustees and their staff decide on the assumptions while the actuaries run the numbers to determine what outcomes those assumptions would produce. In practice, however, the actuaries have a very strong influence on the assumptions – both the "headline" assumptions as well as numerous sub-assumptions needed to model the Social Security program. The actuaries first present their own preferred assumptions, that the working group takes as a starting point for their own discussions. The actuaries' assumptions can and are changed, but they are not a silent voice in the process. This has both pros and cons. On the pro side, the "political" influence on the Trustees Report is very small compared to reports from any other executive agency. This lends (or should lend) credibility to their findings. On the other hand, it is intended that the Trustees have a strong role in producing the Report, and it can be questioned whether that role is strong enough.


Read more!

Friday, March 21, 2008

How dependent are retirees on Social Security?

Warning: This post is long.

In discussions of the role Social Security plays in providing retirement income, one will often hear some variant of the following: Social Security

is the major source of income (providing 50% or more of total income) for 66% of the beneficiaries. It contributes 90% or more of income for one-third of the beneficiaries and is the only source of income for 22% of them.”

These are official SSA statistics. Perceptions of Americans' dependence on the Social Security program help influence views regarding the shape of possible reforms. For that reason, as well as others, it is important to have a clear idea what these statistics mean.

The interpretation of these statistics, and their sensitivity to alternate formulations, are the subjects of an important series of papers in the Social Security Bulletin by Lynn Fisher, an economist at the Social Security Administration.

Fisher shows that commonly used figures regarding seniors' dependency on Social Security rely on a series of measurement decisions, any of which could reasonably be decided in other ways. Using plausible alternate methodologies, the percentage of seniors entirely dependent on Social Security – around one-in-five, by the standard measure – could be as low as 3.5 percent.

Fisher examines four potential sources of bias in how we measure dependence on Social Security benefits:

  • Unit of measurement: do we count "elderly units" or individuals?
  • Benefit reporting: do we use self-reported benefit levels or rely on government data?
  • Asset income: do we include only regular income payments, irregular payments such as IRA or 401(k) withdrawals, or even assets that can be liquidated to produce income?
  • Non-cash benefits: Should we include non-cash benefits such as energy, food or housing assistance?

Any number of reasonable answers can be made to these questions. What is important is that people understand these choices when they ask "How dependent are retirees on Social Security?"

Unit of measurement: The SSA measures of dependence are expressed in terms of “aged units.” Aged units treat each marital unit (married couple or nonmarried individual) as one unit. A non-married individual has only his or her own income and demographic attributes.

How can this affect measured levels of dependence on Social Security? In two ways. First, single individuals tend to have lower incomes, and therefore be more dependent on Social Security, than do married couples. However, since both a single individual and a married couple count as one unit, this can overstate the percentage of individuals who are dependent on Social Security. For instance, if a single person was entirely dependent on Social Security while a married couple was not, on a ‘aged unit’ basis 50% would be wholly dependent on Social Security while on an individual basis only 33% would be.

Second, a non-married individual may share a household with other individuals, but the aged unit does not include the resources of these non-married cohabitants. (Thus, the ‘aged unit’ measure is not as broad as a ‘household’ measure.) If non-married cohabitants share incomes and costs, this can cause overstatement of unmarried individuals’ dependence on Social Security.

Using the individual as the unit of reporting and assuming that family income is shared, the percentage of seniors wholly dependent on Social Security drops from around one-fifth to around 13%.

Asset income: As the pension world shifts from traditional defined benefit plans to defined contribution plans, in which individuals would draw down their account balances to fund retirement expenses, one would expect that the share of seniors reporting asset income would increase. The measured percentage has actually decreased from 1991-2000, but this may be due to the limitations of the CPS survey data SSA uses in its calculations of Social Security dependency. Fisher turns to another survey – the Federal Reserve’s Survey of Consumer Finances, which emphasizes measures of asset holdings, to supplement existing data.

Fisher found that SCF data supported the view that receipt of asset income had remained roughly constant from 1991-2000. From this improved measure of asset holdings, she was able to infer receipt of asset income. This previously unreported asset income was relatively small, but concentrated among lower earners. While it does not greatly affect the average level of dependence on Social Security, it would lower the percentage wholly reliant on the program, from around 20% to around 10%.

Survey data: Fisher examines two issues dealing with data. First, how the Census Bureau’s Current Population Survey (CPS), which SSA uses to calculate its dependency statistics, compares to the Survey of Income and Program Participation (SIPP), another Census survey that can be used to calculate the income of the aged. Second, Fisher examines how results differ when survey data regarding receipt of Social Security benefits is replaced with administrative data.

Survey choice: The SSA dependency data are derived from the Census Bureau’s Current Population Survey (CPS). One advantage of the CPS is that a new survey is conducted annually, allowing for more up-to-date data. Another Census survey, the Survey of Income and Program Participation (SIPP), is conducted less frequently but asks more detailed questions. Survey subjects are asked about 70 sources of income, versus 35 in the CPS, and the survey takers check back with subjects four times per year, versus only once in the CPS. The SIPP also asks detailed questions about financial assets, while the CPS does not.

Administrative data: SSA’s figures regarding dependency on Social Security use self-reported levels of Social Security benefits contained in the CPS survey. However, researchers have matched CPS survey results to SSA administrative files to see how accurately individuals can recount their Social Security benefits. For a number of reasons, individuals misreport their Social Security benefits – confusing them with SSI or other benefits; reporting them net of Medicare Part B premiums, etc.

Taken together, using both the SIPP survey data and matching survey findings to administrative data and reduce reported dependence on Social Security benefits. Using 1996 survey data, Fisher found that the percentage 100% dependent on Social Security using the CPS with self-reported benefits was 17.9%; add administrative data to the CPS and dependence fell to 17.3%; use the SIPP survey with self-reported benefits and 100% dependence fell to 8.5%, and using SIPP matched to administrative records dependence fell to 8.4%.

Effects in combination: Fisher examines a number of different methodological questions separately, the sees how they affect measured dependence on Social Security when used in combination. Table 1 provides details: Combining all the issues discussed above and applied to 1996 data, the percentage of individuals over 65 depending on Social Security for 100% of their income declined from 17.9% to 4.8%. The percentage depending on Social Security for 90% of their income declined from 30.4% to 13.7%.

In an appendix, Fisher explores the effects of including non-cash benefits, such as food stamps or housing or energy assistance. These are not cash, but are nevertheless valuable resources. If non-cash benefits are included, the percentage wholly reliant on Social Security declines to 3.5%.

As we consider potential reforms to the Social Security program, it is important to have good information regarding the retirement incomes of current seniors, and for policymakers to understand what existing information really means. Put another way, if someone asks the qualitative question, "What percentage of seniors are totally dependent on Social Security?", the answer can range from as high as 20% to under 4%. The two different answers may well lead to two different policy conclusions.

Read more!

Wednesday, March 19, 2008

New paper: "How the Income Tax Treatment of Saving and Social Security Benefits May Affect Boomers’ Retirement Incomes"

The Urban Institute has released a new paper by Barbara A. Butrica, Karen E. Smith, and Eric J. Toder analyzing how four potential changes to tax law would affect retirement income. The changes simulated are:

  • a) reducing contribution limits on 401(k) plans;
  • b) extend tax cuts for capital gains and dividend income;
  • c) index thresholds for income taxation of Social Security benefits; and
  • d) Eliminate Social Security thresholds and include 85% of Social Security benefits in adjusted gross income
The authors conclude that indexing thresholds for income taxation of Social Security benefits would have the largest effects on the middle class. These individuals do not generally reach 401(k) contribution limits, so lowering them would have a smaller effect than on higher earning workers. Likewise, the middle class tends to have lower asset income relative to other income, so extending cuts in capital gains and dividend taxes would have a smaller effect. Eliminating Social Security tax thresholds would have the largest negative effect on middle earners. Read more!

New paper: "Can Faster Economic Growth Bail Out Our Retirement Programs?"

Former CBO director Rudy Penner, a senior Fellow at the Urban Institute, examines how much economic would reduce the burden of rising entitlement costs. The money paragraph on Social Security:

A 0.5 percent increase in the annual rate of productivity growth reduces the 75-year actuarial deficit from 1.95 percent of payroll to 1.39 percent (or by 0.56 percent of payroll). Roughly speaking, it would be necessary to double today’s 1.9 percent assumed rate of productivity growth to solve Social Security’s long-run actuarial problem with economic growth. Although there have been short periods of such high productivity growth over the past five decades, it is totally implausible to believe that such growth could last over 75 years.
The full text is available here.
Read more!

Tuesday, March 18, 2008

From the archives: 2005 presentation on Shiller Social Security paper

While in the employ of the federal government I often did research work that didn't end up seeing the light of day. Most of it deservedly so, but in this case a presentation from back in 2005 may be of interest. At that time, Robert Shiller of Yale wrote a short paper using historical data arguing that President Bush's personal accounts plan would have been a bad deal for most participants.

Shiller's paper was widely cited at the time, but also widely misinterpreted, I believed. Using some very reasonable assumptions, Shiller's paper could be used to show that almost every cohort of personal account participants would have come out far ahead through participating -- an average of around $3,500 per year, in one case. I put together a presentation for the annual conference of the Association for Private Enterprise Education in Orlando. Here I've included the presentation with some explanatory notes.

In any event, the presentation is downloadable here.

Read more!

A really good ad

I love this ad. While in the tradition of ads you see in Washington, DC, where you're not exactly sure what they're selling, the statistic that Hallmark sold 85,000 100th birthday cards last year is a great heads-up to people planning for retirement that there's a non-trivial chance of living a long time. (Oh, now I get it -- Allstate sells annuities...). You can download the ad in pdf form here.

For those wanting more details, I've calculated the likelihoods that at least one member of a 65-year old couple will survive to a given age:

Age 70: 99.27%
Age 75: 95.83%
Age 80: 86.87%
Age 85: 69.15%
Age 90: 43.46%
Age 95: 18.74%
Age 100: 4.66%.

This is an area where I and many of those who have favored Social Security personal accounts were -- how shall I put this? -- wrong. While there is no reason account holders couldn't purchase annuities at retirement, one attribute of personal accounts that was often promoted was the ability to "unwind" the Social Security annuity to allow for periodic withdrawals or bequests. Annuities provide very valuable insurance against outliving your assets -- a standard finding is that you'd need around $150,000 in a world without annuities to provide the income security of an annuity costing $100,000. While accounts have a number of other strong points, unwinding the Social Security annuity isn't really one of them.
Read more!

Allan Sloan on the Trust Fund

In today's Washington Post, business columnist Allan Sloan argues that the Social Security trust fund "has no financial value." Here's the money paragraph:

Say that Social Security calls the Treasury sometime in 2017 and says it needs to cash in $20 billion of securities to cover benefit checks. The only way for the Treasury to get that money is for the rest of the government to spend $20 billion less than it otherwise would (fat chance!), collect more in taxes (ditto), or borrow $20 billion more (which is what would happen). The spend-less, collect-more, and borrow-more options are exactly what they would be if there were no trust fund. Thus, the trust fund doesn't make it any easier for the government to cover Social Security's cash shortfalls than if there were no trust fund.
I've used words to these effect many times, and I believe Sloan's conclusion is substantively correct. But the underlying argument is a lot more complex.

The trust fund isn't of no value because the rest of the government must raise taxes or cut other programs to repay it. That would be the case with any government bond, which leads to absurd conclusions.

The problem with the trust fund that is distinct from ordinary government bonds is that a) when the government borrows from Social Security this borrowing isn't usually counted as part of the budget deficits; and b) debts owed to Social Security aren't usually counted as part of the government's debt.

What do I mean by "usually"? In most cases, the budget deficit numbers you read about are for the "unified budget," that is, the budget including Social Security. If Social Security runs a surplus of, say, $100 billion, this reduces the unified budget deficit by that mount. Likewise, the government debt figure you most commonly read about is the "publicly held debt," that is, debt held by individual investors, Wall Street, foreign central banks, etc. Borrowing from Social Security -- "intragovernmental debt" -- usually isn't included in this amount.

So why does this matter? When borrowing from Social Security isn't counted as "real" borrowing, Congress will tend to borrow more. That is, Social Security surpluses allow the rest of the government to tax less or spend more than it otherwise would have. In this case, the government's total asset position isn't improved, nor is national saving increased. The whole story is outlined in much greater detail here.

So Allan Sloan's conclusion is basically correct, but the way he explains it opens himself to legitimate criticism.

Update: See Dean Baker's post on the same subject here.
Read more!

Monday, March 17, 2008

Are the Trustees too pessimistic?

In the comments, Bruce Webb raises a common argument, that Social Security’s Trustees have generally been too pessimistic about the program's finances. I wrote on this issue a long time ago, and posted my little solvency widget in part to let people play out their own scenarios. But this issue is an evergreen and deserves some more attention.

Bruce argues that the Trustees have been overly pessimistic based on their short-term forecasting record. Bruce compares GDP growth as projected in a Trustees Report with the realized figure one year later. He argues that since 1997 GDP growth has generally been understated and that the Low Cost projections have been more accurate. From here, he concludes that the Low Cost 75-year projections may be more accurate.

A couple points: First, it is hard to draw firm conclusions about the Trustees’ 75-year forecast from the accuracy of their single-year forecasts, since these are very different animals. The SSA Office the Chief Actuary is geared more toward long-term forecasting, while the Congressional Budget Office and private sector forecasters concentrate more on the short-term. Moreover, in many ways long-term forecasts are easier than short-term forecasts, since the long term relies more on underlying fundamentals and less on year-to-year fluctuations. For instance, I cannot say with accuracy whether the weather one week from now will be cooler or hotter than today’s, but I can say with great accuracy that the weather six months from will be hotter simply because of the fundamentals.

Second, Bruce uses one method for assessing the Trustees’ accuracy, but others – perhaps better ones – are available. Here I turn to Chuck Blahous, the number two at the White House National Economic Council and the Bush administration’s chief policy man on Social Security. In a paper presented last year at the American Enterprise Institute, Chuck applied a rigorous test to the Trustees accuracy.

First, he devised two benchmarks of Social Security’s financing health: a), the annual cash balance (surplus or deficit) as a percentage of payroll; and b), the trust fund ratio – the trust fund balance as a percentage of annual costs. At any given time, these are good measures of how well Social Security is funded, clearly more so than GDP growth alone. In 2005, the last year for which data was available when the paper was being written, the annual Social Security cash surplus was 1.55% of taxable wages, and the Trust Fund ratio was 318%.

Next, Chuck looked back over the previous Trustees Reports ranging back to 1983 and analyzed how accurately they projected these outcomes for 2005. Specifically, he asked which set of projections – Low Cost, High Cost, or Intermediate – was the most accurate in projecting the cash balance and trust fund ratio for 2005.

Figure 1 details the Trustees projections of the annual cash flows. Of the 22 Reports, the Low Cost was most accurate in only 4, the Intermediate Cost in 7, and the High Cost in between 8 and 11 Reports (this is due to changes in how the Reports were structured).

Figure 2 repeats the process for the Trust Fund ratio. Here the Intermediate Cost projections have been most accurate, being the best projections in 12 of 22 Reports. The Low Cost projections were most accurate in 6 years, and the High Cost in 4 years.

Clearly other tests could be applied. But one test – summing the squared errors of the different methods, then discounting to control for the difficulty of prediction as time increases – shows the summed errors for the Intermediate Cost projections were only one-fourth the size of those of the Low or High Cost scenarios.

Chuck summarizes the results in this way:

A reasonable summary of the 1983-2004 projections’ accuracy is that the Trustees’ Intermediate projections most accurately predicted the cumulative value of the Social Security surpluses over the totality of the 1983-2004 period. With respect to predicting the balance of annual system operations in 2005, the High-Cost projections were the most accurate in the highest number of reports, although the Intermediate projections exhibited the lowest total error. On balance, by both standards taken together, the Intermediate projections have been the most accurate, and the Low-Cost projections have been the least. Though this does not necessarily mean that the superiority to date of the Intermediate projections is significant for the future, the data clearly contradict the widespread misconception that the Low-Cost projections have in the past been best.

Anyone considering the argument that the Trustees are overly pessimistic must account for the arguments and evidence Chuck Blahous presents in this paper.

In addition to the paper itself, PowerPoints from Chuck and discussants Steve Goss (SSA) and John Sabelhaus (CBO) and video of the entire event are available here at the AEI website.

Read more!

Pension reform in Chile

The Associated Press reports that the Chilean government is preparing to roll out a significant reform of the country's pension program, which after its introduction in 1980 served as model for reform in Latin America and elsewhere in the world, and has been proposed as a reform model in the U.S. Given this, it makes sense to some of the issues involved with Chilean pension reform as they relate to the U.S.

The biggest problem with the personal accounts pension system in Chile, as with the traditional defined benefit program that preceded it, is incomplete coverage. In the U.S. coverage under Social Security is practically universal (less than 5% of workers aren't covered), and the vast majority of those not covered under Social Security take part in another pension program. In Chile, by contrast, the self-employed, who have never been mandated to take part in their pension program, make up almost 40% of the workforce.

For that reason, even though average replacement rates for full-time workers under the Chilean pension plan are quite high -- around 85% percent, versus around 45% under U.S. Social Security -- a large number of self-employed workers would reach retirement with insufficient savings.

In response, the Chilean government has maintained a minimum pension benefit for individuals with 20 years of participation in the accounts system. The minimum benefit is equal to around 25% of the average wage. In U.S. terms, this would make for a minimum Social Security benefit of around $10,500. (Using the Gemini model, I estimated the benefits for 1941 birth cohort, individuals who would turn 67 in 2008. Among those who had claimed benefits, 67% had an initial benefit lower than $10,500. This gives some perspective on how relatively high the Chilean minimum is.)

Since the Chilean minimum pension is means-tested, lower-wage workers have very little incentive to participate in the personal accounts system once they have reached 20 years of coverage. The minimum pension is relatively generous so they would lose much or all of what they saved in their accounts. It is almost certain that the presence of the minimum pension -- designed to assist low-wage workers -- has aggravated the problem of low participation.

What are the lessons of this for U.S. Social Security reform? Probably quite few. No existing reform plans allow workers to effectively opt-out of Social Security, as the Chilean system allows self-employed workers to do. As a result, even in personal account plans that contain a guarantee of current law scheduled benefits, this type of moral hazard problem would not express itself. (There may be other moral hazard problems with such plans, such as the incentive to invest in risky assets, but that is a separate issue.)

For a good summary of the Chilean system and its strengths and problems, see this paper from Mauricio Soto of the Center for Retirement Research at Boston College.
Read more!

Saturday, March 15, 2008

What does it mean to “save the surplus”?

In the comments, Bruce Webb reacts to a line in the Wall Street Journal op-ed:

The whole notion that somehow Congress has been 'raiding', 'looting' or otherwise doing something nefarious with the Trust Fund derives entirely from the implication that the Special Treasuries are not in fact hard investments, which is to say one or more versions of the 'phony IOU' argument.
Well I am not buying. I believe in the Full Faith and Credit of the United States, further I have a pretty good idea of the political firestorm that would erupt at any suggestion that the General Fund as the representative of taxpayers at large would refuse to replay legal obligations to that subset of taxpayers that paid for all those surpluses to start with.

This raises the general question of what it means to "save the surplus" or "raid the trust fund." There are different definitions, and often disagreements over these issues arise from definitional, not empirical, differences.

Orszag and Stiglitz make a useful distinction between "narrow" and "broad" prefunding of pension benefits:

"Prefunding" can be used in a narrow or broad sense. In its narrow sense, prefunding means that the pension system is accumulating assets against future projected payments. In a broader sense, however, prefunding means increasing national saving.

Put in the Social Security context, it's possible to break the narrow versus broad prefunding question down even further. "Saving the surplus" can mean that a dollar of surplus Social Security taxes implies:

  • A dollar increase in the Social Security trust fund: This is the narrowest definition of pre-funding, and in this sense the surplus is indisputably "saved." Any surplus taxes are by law used to purchase special issue Treasury bonds. These bonds carry a market rate of interest and are backed by the full faith and credit of the U.S. government. There is no possibility that the government will not honor these bonds, and there are no reform plans that propose that they not be honored.
  • A dollar increase in the overall budget balance: This is an intermediate level of pre-funding, and most advocates would be satisfied if this level were achieved. If Social Security's cash balance improves by one dollar and nothing else changes in the rest of the budget, then the overall budget balance will improve by one dollar. Borrowing from the public will be reduced by one dollars (or, if the budget were in surplus, one dollar of existing debt could be repaid). At the least, this level of prefunding makes it easier for the government to repay the Social Security trust fund in the future, since the smaller government debt implies lower annual interest costs.
  • A dollar increase in national saving: If an additional dollar of Social Security surplus adds to government saving, by the above process, and if individuals do not alter their saving behavior, then total saving in the economy will increase by one dollar. This saving adds to the stock of investment capital, such as factories, computers, etc., and this additional capital makes future workers more productive and increases economic output. This increased economic output makes it easier to repay the trust fund in the future: wages will be higher and thus tax receipts will be higher even with a constant tax rate. Thus, we could repay the trust fund without making future workers' after-tax wages lower than they otherwise would have been.

When we talk about "saving the surplus" it makes sense to be clear which definition of saving we're relying on. In the WSJ op-ed, I was implicitly referencing the second definition of prefunding: that an increase in the Social Security annual surplus, as would be produced by Senator Obama's proposal, would translate to an equal improvement in the overall budget balance, and therefore a reduction in government borrowing.

But is this likely to take place? A trio of econometric studies by well-respected economists have concluded that Social Security surpluses since the 1980s have not translated to improved budget balances. The basic analytical technique is to ask how changes in the Social Security balance correlated with changes to the overall budget balance, after adjusting for other factors. Kent Smetters of the Wharton Schol, who wrote the first such study, concludes:

"…there is no empirical evidence supporting the claim that trust fund assets have reduced the level of debt held by the public. In fact, the evidence suggests just the opposite: trust fund assets have probably increased the level of debt held by the public."

Barry Bosworth and Gary Burtless of the Brookings Institution, using a sample of OECD countries to supplement results focusing on the U.S., conclude:

"A large portion of the accumulation within national social insurance systems is offset for the government sector as a whole by larger deficits in other budgetary accounts. On average, OECD countries have been able to save only a small portion of any funds accumulated within their social insurance systems in anticipation of large expected liabilities when a growing fraction of the national population is retired. Between 60 and 100 percent of the saving within pension funds is offset by reductions in government saving elsewhere in the public budget."

In other words, a dollar of Social Security surpluses tends to be offset by 60 cents to one dollar in increased spending or reduced taxes in the non-Social Security portion of the budget.

John Shoven of Stanford and Sita Nataraj of Occidental College examined trust fund saving throughout the federal budget. Their conclusions are summarized as:

"The authors find a strong negative relationship between the surpluses: an additional dollar of surplus in the trust funds is associated with a $1.50 decrease in the federal funds surplus. This finding is not significantly different from a $1.00 decrease, which would suggest a dollar-for-dollar offset of trust fund surplus with spending increases or tax cuts; the authors are able to reject the hypothesis that the full dollar of trust fund surplus is saved by the government."

To sum up, the best evidence suggests that Social Security surpluses, rather than building savings to help pay future Social Security benefits, instead tend to subsidize present consumption. Put another way, Social Security surpluses allows current spending to be higher, or current taxes lower, than they otherwise would be. Several preliminary policy conclusions follow from these findings:

First, that it makes little attempt to increase Social Security surpluses in the near-term, unless a more reliable mechanism to save these surpluses is found. Investment in assets other than Treasury bonds might provide better prospects for broad saving; personal accounts also might do so. But these are questions for a different post.

Second, changes to Social Security taxes and benefits prior to the trust fund exhaustion date (currently 2040) can be justified. If surpluses since the 1980s had been "saved" in a broad budgetary sense, one could say that taxpayers during those years had fully "paid for" their future benefits, even if trust fund repayment exerted a burden on the non-Social Security budget in the years from 2017-2040. However, if Social Security surpluses were "spent" this implies that taxpayers since the 1980s enjoyed a higher level of government services or lower level of government taxes than would otherwise have been the case. Given this, their moral claim that benefits cannot be changed prior to the trust fund's exhaustion appears weaker.

Read more!

Thursday, March 13, 2008

Mark Thoma on John Shoven and raising the retirement age

Stanford economist John Shoven argues for increasing the Social Security retirement age, based on a novel redefinition of the term “age”: for Shoven, age should be measured in terms of years from death rather than years from birth. The Social Security retirement age should rise in years-from-birth terms, Shoven says, given that years from death are increasing as mortality declines. Shoven says:

Just consider the consequences of altering the age when entitlement benefits kick in or retirement becomes mandatory to these new inflation-adjusted measurements. It doesn't mean shortening retirements, just stabilizing them. In 20th-century America, the average length of retirement grew from two years to more than 19 years. As life expectancies continue to rise, retirements will continue to get longer -- and the pension bill far larger. If benefits and retirements are governed by mortality risk instead of age, the costs will be far more manageable. We've witnessed dramatic improvements in life expectancies over the past century. It's time we dramatically improve the way we measure age as well.

Mark Thoma disagrees. First, he says:

there is no guarantee that the ability to work and lead a productive life expands at the same rate as life expectancy (e.g., if most of the extension in life expectancy in the future comes from expensive interventions toward the end of life rather than improvements in health during, say, the late 60s, that make working easier then there would be no reason to extend the retirement age). It would be better to define retirement in terms of the minimum of these two concepts, the time at which the typical person can no longer be expected to work full-time on a typical job that may have physical demands, and the life expectancy adjusted retirement age discussed above.

Of course, there are no “guarantees” that the ability to work rises commensurately with expected age of death, but this seems a remarkably high standard of evidence to demand. The evidence seems to indicate that health status is improving with longevity – that is, we are not simply surviving a few more years in decrepitude but the process of aging is in fact moving more slowly.

For instance, John Turner examines health status by age and ethnicity among individuals aged 50 to 64. In all categories, the percentage of individuals reporting fair or poor health status has declined significantly since 1982.

The percentage of individuals with physically demanding jobs has also declined: from 1950 to 1992, the percentage of individuals reporting jobs requiring frequent lifting of heavy objects has fallen by from 20 to 8 percent. Turner concludes that “it appears clear that if demand for older workers were sufficient, it would be feasible to raise the Early Eligibility Age for Social Security to 63.5 in order to promote longer worklives.”

Thoma’s second objection to Shoven’s argument is “that Social Security is not the entitlement problem we should worry about, that title belongs to Medicare where costs are expected to increase rapidly in the future.”

This argument is flawed, for two reasons. First, simply because one problem is larger than another does not mean that we address only the largest one. Social Security is the largest program of the federal government and its costs are projected to rise by over 20% relative to its tax base in the next 10 years, with further increases in following decades. Given that reform is essentially a problem of smoothing cost burdens over different cohorts, it’s not clear how delaying action makes for a better outcome.

Second, Social Security is a more mature policy issue than health care. We know fairly precisely what the cost drivers are for Social Security, while for health care the components of excess cost growth are less clear cut. Moreover, for Social Security we know the range of options fairly well, making a potential compromise easier to envision. On the health care side it’s not well understood even the degree to which we should want to restrain cost growth, much less the most efficient ways to do so.

Read more!

Wednesday, March 12, 2008

Obama's Social Security Proposal

A piece from me in today's Wall Street Journal:

The Obama Tax Hike

Until recently, Sen. Barack Obama took a responsible position on Social Security, noting the urgency of reform and saying all options should be on the table.

But having cornered himself among Democratic activists whose attitudes toward Social Security reform range from demagoguery to denial, Mr. Obama has recently veered sharply left. He now proposes to solve the looming Social Security shortfall exclusively with higher taxes.

"Once people are making over $200,000 to $250,000," Mr. Obama says, "they can afford to pay a little more in payroll tax." No shared sacrifice, no outreach to moderates or conservatives, here.

Mr. Obama's proposal is to make a significant change to the payroll tax system. Currently, all wages below about $100,000 are subject to a 12.4% Social Security payroll tax. But all wages above that amount are not subject to the tax. Mr. Obama wants to eliminate the cap, but, in a concession to taxpayers, exempt wages between $100,000 and $200,000. He wants to create a "donut hole" in the taxing mechanism that pays for the nation's largest retirement program.

The problem is two-fold: His proposal would be a very large tax hike, yet it won't be enough.

Mr. Obama's plan fixes less than half of Social Security's long-term deficit, making further tax increases inevitable. The Policy Simulation Group's Gemini model estimates that Mr. Obama's proposal, if phased as Mr. Obama suggests, would solve only part of the problem. A 10 year phase-in, for example, would address only 43% of Social Security's 75-year shortfall. And this is assuming that Congress would save the surplus from the tax increases -- almost $600 billion over 10 years -- rather than spending it, as Congress does now.

What's more, Mr. Obama's plan would keep Social Security in the black for only three additional years. Under his proposal, annual deficits would hit in 2020, instead of 2017. By the 2030s the system would still run an annual deficit exceeding $150 billion.

Mr. Obama's modest improvements to Social Security's financing come at a steep cost. The top marginal federal tax rates would effectively increase to 50.3% from 37.9%, equivalent to repealing the Bush income tax cuts almost three times over.

If one accounts for behavioral responses, even the modest budgetary improvements from Mr. Obama's plan are likely to be overstated. If employers reduce wages to cover their increased payroll-tax liabilities, these wages would no longer be subject to state or federal income taxes, or Medicare taxes. A 2006 study by Harvard economist and Obama adviser Jeffrey Liebman concluded that roughly 20% of revenue increases from raising the tax cap would be offset by declining non-Social Security taxes. Assuming modest negative behavioral responses, Mr. Liebman projected an additional 30% reduction in net revenues, leaving barely half the intended revenue intact.

Mr. Obama's plan would also dramatically raise incentives for tax evasion, further degrading revenue gains. Many high-earning individuals evade the Medicare payroll tax by setting up "S Corporations," paying themselves in untaxed dividends rather than taxable wages. John Edwards avoided $590,000 in Medicare taxes this way in the 1990s. Under Mr. Obama's plan, Mr. Edwards's savings would have exceeded $3 million. With that much at stake, the incentive to follow Mr. Edwards lead will be that much greater.

Mr. Obama's plan shows the limits to taxing the rich as a solution to Social Security's problems. Top earners would effectively be tapped out, with taxes as high as economically and politically feasible, yet most of Social Security's deficit, and the much larger shortfalls in Medicare, would remain.

The U.S. already collects far more Social Security taxes from high earners than other countries do. Social Security taxes here are currently capped at about three times the national average wage -- far above other developed countries. In Canada and France payroll taxes are levied only up to the average wage. In the United Kingdom, taxes stop at 1.15 times the average wage; in Germany and Japan at 1.5 times. Social Security is already more progressive than these countries' pension programs, and Mr. Obama's plan would make it more so.

President Bill Clinton considered lifting the wage ceiling modestly, but was skeptical of eliminating it outright. Doing so would "tremendously change the whole Social Security system . . . We should be very careful before we get out of the idea that this is something that we do together as a nation and there is at least some correlation between what we put in and what we get out," Mr. Clinton said in 1998. "You can say, well, they owe it to society. But these people also pay higher income taxes and the rates are still pretty progressive for people in very high rates."

Social Security's shortfalls are primarily attributable to society-wide trends of lower birth rates and longer lifespans. If we want to retain the shared character that underpins its political support and distinguishes it from traditional welfare programs, we need to share the burdens of reform proportionately. Mr. Obama should drop his exclusive focus on raising taxes and return to his previous view, that Social Security faces significant problems requiring prompt attention. All options should be on the table.

Mr. Biggs, a former principal deputy commissioner at the Social Security Administration, is a resident fellow at the American Enterprise Institute.

Commentary on the op-ed from:

Donald Luskin

Ryan Ellis

Ankle Biting Pundits Read more!

Friday, March 7, 2008

CRS Report on Social Security Transition Costs

Several blogs are reporting an update (or perhaps simply the public release) of a 2007 Congressional Research Service policy paper on the transition costs associated with Social Security personal accounts. While any exploration of this tricky issue is welcome, this study – perhaps due to a desire to produce quantifiable results with limited data – unfortunately mischaracterizes the question.

“Transition costs” arise when a pay-as-you-go program is converted in part or whole to a pre-funded program. In a pay-as-you-go system today’s retirees are supported by today’s workers; in a funded system today’s retirees are supported by their own savings, accumulated during their working years.

Any effort to prefund future benefits involves transition costs, be it through “carve out” accounts funded out of the existing payroll tax, “add on” accounts funded on top of the payroll tax, or through investment of the Social Security trust fund in assets other than Treasury bonds. (I will discuss this latter case in a separate post).

But for simplicity, let’s stick to the common example where today’s workers invest part of their payroll taxes in personal accounts. Since Social Security is a pay-as-you-go program, today’s taxes are earmarked to pay today’s benefits. If part of those taxes are invested to pay tomorrow’s benefits, the system face a temporary shortfall. This shortfall, which declines and eventually disappears as workers with accounts begin to retire, is referred to as the transition cost. Intuitively, the larger the amount of prefunding – that is, the larger the accounts and the higher the share of workers who participate – the larger the transition cost.

Transition costs bring with them transition benefits. When workers with accounts retire, the tax rate needed to provide a given level of total benefits declines since part of those benefits are provided by the accounts, which were funded in the past.

Where things get difficult, and where the CRS report runs into problems, is when we attempt to quantify the transition costs of a given reform plan and compare to other plans. The CRS report defines transition costs as “the dollars put into the IA (Individual Account) by the government [minus] any dollars taken from the IA to help pay benefits over the 75-year actuarial period.”

While this definition sounds reasonable, it encounters several significant problems.

First, in any measure that consistently counts both the contributions to accounts and benefits paid from accounts (e.g., closed group or infinite horizon measures), the measured transition costs for any given cohort of participants will be negative so long as the accounts’ assumed returns exceeds those of the Social Security trust fund. This is simply because the account contributions would be compounded forward at a higher interest rate than they are discounted back at. This is unnecessarily generous to personal accounts: if accounts earn more than the riskless interest rate it is because they invest in riskier assets. In any case, it's not clear how netting benefits paid against account contributions relates to the intuitive view of transition costs.

Moreover, the net transition costs of reform plans can differ based upon the assumed interest rate earned by their personal accounts. Accounts with aggressive investment portfolios will pay more benefits and thus have lower transition costs than accounts with more conservative investments. The accounts in Kolbe-Stenholm, for instance, are invested in 50% stocks and 50% long-term Treasury bonds. With assumed returns of 6.5% and 3.0% and administrative costs of 0.3% of assets, the net projected return is 4.45%. The Ryan plan, on the other hand, assumes 65% investment in stocks and 35% in corporate bonds with administrative costs of 0.25%, for a projected net return of 5.2%. As a result, the Ryan plan has among the lowest transition costs and Kolbe-Stenholm the highest, despite the fact that Ryan’s accounts are twice as large as Kolbe-Stenholm’s. This conclusion simply seems wrong. (There may be other plan-to-plan measurement issues as well: it appears that for some plans the study counts the actual benefits paid out of personal accounts, while for others it counts the reductions in traditional benefits due to account participation. While related, these are not the same thing. In particular, cuts in traditional benefits for personal account holders are larger in plans with guarantees of receiving at least scheduled benefits. Without these guarantees most workers would not accept accounts on the terms presented. If so, however, then the cost of the guarantee would need to included.)

Second, countering the above methodological generosity is the use of a 75-year measurement period to calculate transition costs. In fact, the only reason any of the reform plans show any significant transition costs is due to measurement using a truncated 75-year period. Net transition costs are driven almost entirely by account contributions made during the 75-year period that are paid after the 75th year. While one can argue about measuring system financing with a 75-year period or the infinite horizon, in this case the 75-year horizon seriously skews the results. And, as above, it’s not clear how this measure relates to the intuitive view of transition costs outlined above.

The problems presented above don’t imply that measuring transition costs is easy. The President’s Commission to Strengthen Social Security presented a measure of “transition financing” that stated:

In every year where financing needs are greater under the Reform Model than they would be under current law, that year is identified as a “transition year.” All required extra financing is added up for each transition year, and the sum is given on the table.

Under this definition, a reduction in the Social Security surplus due to personal account is not counted as part of a transition cost. In addition, if reductions in traditional benefit growth reduce the need for overall financing to the program, this also reduces measured transition costs. Again, it’s not clear that this measure hits on the intuitive notion of transition costs.

Read more!