New paper: “Stock Market Fluctuations and Retiree Income: An Update”

Gary Burtless from the Brookings Institution has released an update to his earlier work on how stock market volatility could affect individuals holding Social Security personal accounts. Here's the summary, followed by my comments:

The recent plunge in home values and even bigger dive in stock prices offer painful reminders of why Social Security seemed like such a good idea in the 1930s. Benefits are predictable, are guaranteed by the government, and are adjusted every year to keep their purchasing power stable. In contrast, workers who count on the stock market to fund their retirement have seen their savings shrink more than 40% over the past year. The question is: what kind of retirement plan offers the best guarantee workers will receive a predictable and comfortable income when they grow old?

Luckily for most older Americans, the cornerstone of their retirement income is still a Social Security check. Social Security plays a crucial role in maintaining the incomes of Americans past the age of 65. Last year it accounted for 39% of the total income received by the elderly. It is a particularly important source of income for low-income seniors. For aged Americans in the bottom one-fifth of the income distribution, it accounts for nearly $9 out of every $10 they receive. The benefits and returns are more secure than incomes from private saving accounts, and they are indexed to inflation, which is rarely the case for private pensions or annuities.

Financial planners often recommend that workers aim to replace 75% to 85% of the wages they earn before retirement. Suppose workers set aside 4% of their salaries to meet this goal. How much would their retirement incomes fluctuate, depending on the start and end dates of their careers and the investment strategies they choose? Between 1999 and 2002 both the stock market and the yield on bonds declined. Wage earners who worked for 40 years and invested all their retirement savings in stocks would have seen their wage replacement rates fall 53 percentage points. Workers who followed a more conservative investing philosophy and placed half their retirement savings in bonds would have seen their replacement rates fall 18 percentage points. The latest market turmoil has delivered another jolt to workers who count on their stock investments to pay for retirement. Between October 31, 2007, and October 24, 2008, the drop in stock prices has reduced the expected retirement income flow from a stock-invested savings account by 46%.

The Great Depression and the stagflation years from 1974-1983 were the most recent periods of great economic uncertainty. Few people in those years suffered under the illusion that a private savings account offers a secure foundation for a comfortable retirement. Big selloffs in the stock and bond markets persuaded most Americans that government-guaranteed pensions were valuable and well worth preserving. The recent market selloff may have the same salutary effect on voters' opinions. Social Security's problems still need fixing. But it is hard to argue that the most sensible fix will involve scaling back Social Security's basic promises in order to make room for a bigger private savings system.

Since I'll disagree with Gary regarding some of the details, let me state upfront that I agree with many of his qualitative points and many of his conclusions. The stock market is volatile, Social Security accounts don't pay higher returns when adjusted for risk, and the recent market fluctuations do show the value of a combined defined benefit/defined contribution provision of pension income. In some senses, the numbers I'll present -- while a better representation of what an actual reform plan would have paid -- look too good; future returns could be lower and more volatile than in the past, so beware that past performance doesn't guarantee future results.

I have written on this topic recently and have a longer paper coming out soon, but I'll try to explain here what drives the differences between Gary's results and my own. While I don't question his calculations, I believe his modeling of the accounts differs in significant ways from the structure of actual account based reform plans and these modeling differences account for much of his results.

Let me first explain the stylized personal account plan that I modeled:

• Individuals would invest 4 percentage points of the 12.4 percent social security tax in a personal account. At retirement, the account would be annuitized to produce a steady monthly income.
  
• The account would hold a "life cycle" fund that holds 85 percent stocks through age 29, declining to 15 percent stocks by age 60. Individuals are assumed to work from 21 through 64 and retire at age 65.
  
• In exchange for being able to divert part of their payroll taxes to a personal account, individuals would give up traditional benefits equal to their account contributions compounded at the government bond rate of return. This "shadow account" would be annuitized on the same terms as the real personal account, and the annuity payment would be deducted from the worker's traditional benefits. This shadow account is designed to keep the trust fund roughly neutral with regard to personal account participation.
  

To keep things simple, total Social Security benefits would increase to the degree that the return on the real personal account exceeded the return on the all-government bond accounts.

I simulated personal accounts using stock and bond returns from 1871 through September 2008. The chart below shows total Social Security benefits – traditional benefit, minus shadow account offset, plus personal account annuity – relative to pre-retirement earnings. (Note that my replacement rates are relative to the workers Average Indexed Monthly earnings while Gary's are to earnings during the worker's 50s, so the levels of replacement rates will differ. However, the volatility is what we're interested in here so it's not a huge deal.)

The baseline replacement rate for Social Security benefits alone is 39 percent; this is the replacement rate an individual would receive if he did not participate in an account. The average total Social Security replacement rate for an account holder is 45 percent. The interquartile range is from 44 percent through 46 percent, meaning that half the cohorts receive replacement rates in this range. The minimum replacement rate was 41 percent while the maximum replacement rate was 48 percent.

I'll leave it to others to judge whether this constitutes too much volatility in retirement income derived from Social Security. However, it's clear that this level of volatility is significantly lower than found in Gary's paper. Why? Three reasons:

• First, I am modeling replacement rates from the total Social Security benefit, while Gary models only replacement rates from the account alone. Because of the interactions between the account and the shadow account, detailed below, I believe this is the only way to realistically analyze volatility of retirement income from reform proposals that could actually be considered.
  
• Second, part of the volatility in replacement rates in Gary's simulation derives not from risky stock returns but from differences in interest rates at the time the account is converted to an annuity. The annuity payment rises or falls based on the interest rate at the time the annuity is purchased. In the stylized reform plan I simulate – which again, more closely resembles actual reform plans than does Gary's approach – both the annuity from the personal account and the offset from the shadow account are calculated using the same interest rate. So if interest rates happen to be high at the time you retire, your personal account annuity would be higher, but so would the reduction in your traditional Social Security benefit. Likewise, if interest rates were low, both the account annuity and the traditional benefit offset would be lower. By itself, this eliminates a lot of risk.
  
• Third, Gary's portfolio is 100 percent stocks, while mine – following the default portfolios in most reform plans – is around 53 percent stocks on average. As a result, replacement rates would vary more in Gary's simulation than mine. Had I used an all-stock portfolio, the average replacement rate would have risen to 65 percent, with an interquartile rate of 53 to 72 percent. The minimum and maximum replacement rates were 42 percent and 108 percent. Again, people can judge for themselves whether the higher mean replacement rates justify the increased range of outcomes.
  

The point here is not to dispute Gary's numbers, which I'm sure are correct, nor is it to argue that we should all go out and put all our Social Security money into the stock market. However, if we more realistically simulate a reform plan as it might actually be introduced in Congress, the historical results look significantly better what Gary's paper might suggest. I have my data and calculations in an Excel spreadsheet, which I'm happy to share if anyone is interested.

Update: U.S. New's Emily Brandon discusses the two papers here.