As the political battles over taxes and the deficit rage on, both Democrats and Republicans seem willing to consider reforming the federal income tax code. Such reform generally calls for paring back tax preferences with lower tax rates as part of the bargain. One proposal currently under discussion is limiting lifetime accumulations in tax-preferred pension and retirement savings plans.

Throughout much of the 1980s and into the 1990s, legislative initiatives cut back the amounts workers could save for retirement in employer-sponsored retirement plans and individual retirement accounts (IRAs).1 There are already limits on the earnings against which contributions can be made as well as amounts paid out of defined benefit plans and paid into defined contribution plans.

As in the past, policymakers seem particularly interested in preventing high earners from receiving “too much” benefit from the tax preferences for retirement saving. But the ramifications of such limits vary widely along with interest rate swings and other financial market conditions. This article discusses the proposals to limit lifetime accumulations of tax-preferred retirement savings and shows how these proposals would interact with existing limits. The simulations illustrate that the ultimate effects of such limits on workers would be less predictable, more restrictive and broader than policymakers intend.

Administration proposal to limit lifetime retirement savings

The Obama administration has proposed to limit lifetime savings in tax-qualified retirement plans. The lifetime maximum would be the equivalent accumulation required for a worker at age 62 to finance an annuity of $205,000 a year as of 2013, with the annual income target indexed to the consumer price index (CPI). At conversion rates in effect at the beginning of 2013, the lifetime limit would be $3.4 million. The new limit would create a regulatory equivalence with existing defined benefit pension limits.

While the annuity amount might be roughly predictable, the capital required to purchase it would be a moving target (Figure 1), due to fluctuations in the interest rates used to determine lifetime income streams. For example, in 1982 a 62-year-old could purchase a $205,000 annual annuity for $1.3 million, while five years later, the same annuity costs $2.2 million.

Figure 1. Balances needed to finance a $205,000 annuity at age 62, 1915 – 2013 ($ millions)
Towers Watson Insider - Figure 1. Balances needed to finance a $205,000 annuity at age 62, 1915 – 2013 ($ millions)

Source: We used interest rates from year-end 1913 through year-end 2012 and historical data on U.S. stocks, interest rates and inflation developed by Professor Robert Shiller from Yale University.2 Our data are an extension of the data Shiller used for his book, Irrational Exuberance. We used a contemporary life table so there is no variation in life expectancies in the analysis. 

Interaction of existing constraints with proposed lifetime limit

The 1974 Employee Retirement Income Security Act (ERISA) limited plan contributions and payouts, and legislators have adjusted those limits over the years. In 1982, legislators introduced limits on the compensation upon which contributions to retirement savings plans are based, which have also been modified since then.

The limit on covered compensation, which is indexed to the CPI, was $255,000 in 2013. The maximum annual payout from a defined benefit plan was $205,000 in 2013. The limits apply to single-life annuities beginning between ages 62 and 65, with the benefits adjusted to reflect starting age and payment form. In 2013, the maximum contribution to employer-sponsored defined contribution plans was $51,000. Under another limit, contributions may not exceed a worker’s annual compensation from the employer. Participants in 401(k) and similar plans were limited to maximum regular deferrals of $17,500 in 2013.

Because of increasingly strict limits over the 1980s and 1990s, the maximum fundable defined benefit for workers retiring in the late 1970s and early 1980s was worth 50% more than benefits for those retiring in 2013. For early retirees — those retiring before age 65 — the reductions were even more dramatic. Defined contribution plan participants could contribute more than twice as much before 1982 as they could in 2013.3 Simulation results — assuming an investment allocation of 60% U.S. equities and 40% bonds, with annual rebalancing — show that an older worker nearing retirement today might have a tax-favored balance of slightly more than $3.3 million under historical limits. Had today’s compensation and annual compensation limits applied over this worker’s full career, however, the accumulation would not have exceeded $2.5 million.

Saving for retirement is a long-range process, and the proposed lifetime limits would likely be imposed on a pro rata basis over a worker’s career. To assess the implications of a lifetime cap for a highly paid worker retiring in 2013, we apply the proposed limit retroactively given historical inflation rates, interest rates and financial market returns from 1976 through 2012. The simulation assumes that current compensation and contribution limits were indexed to annual rates of inflation over the period. 

To apply the lifetime limit on an age-62 accumulation throughout a worker’s career, the limit would be discounted back to the worker’s current age. For example, for a 25-year-old worker back in 1976, the equivalent of a $205,000 annuity in 2013 would be $50,000 in nominal terms. The lump sum equivalent of that age-62 annuity would have been just under a half million dollars for a retiree in 1976, but for a 25-year-old worker, the lump sum equivalent would reflect the interval between the worker’s current age and retirement age. Using 1976 interest rates, this worker’s maximum age-25 accumulation would have been discounted to roughly $28,000. In this case, the discounted lifetime accumulation limit would have had no effect on the young worker’s accumulation because the contribution limit would have been $12,000 in 1976. The contribution limit — not the discounted lifetime cap — would have constrained the 1976 accumulation.

Figure 2 shows how the lifetime limit would evolve over this worker’s career under nominal contribution limits versus under a discounted lifetime cap coupled with contribution limits. 

Figure 2. Tax-qualified accumulations from 1976 through 2012 assuming lifetime limit is imposed over entire career versus on a pro rata basis as worker ages
Towers Watson Insider - Figure 2. Tax-qualified accumulations from 1976 through 2012 assuming lifetime limit is imposed over entire career versus on a pro rata basis as worker ages

Source: Developed by the authors as described in the text

The top line in Figure 2 shows the accumulation assuming current contribution limits had reached their present levels through consumer price indexing since 1976. The bottom line is the estimated account balance under both current limits and proposed lifetime limits. The total accumulated balance at age 62 is just over $1.5 million — less than half the proposed lifetime limit. 

The lifetime accumulation is so much less than the proposed limit mostly because of extremely high interest rates in the early 1980s. In discounting, higher interest rates mean lower limits. So this worker’s contributions were either drastically reduced or suspended entirely during high-interest-rate years. If the lifetime cap were the only constraint, the worker could have made up the lost savings time with catch-up contributions. But he or she could not make up the shortfall because of the annual contribution limits.

To ensure that Figure 2 doesn’t represent an exception, we simulate the effects of the current limits alone and combined with the lifetime cap for 100 unique career-length periods. In the first period, the worker started saving in 1877 at age 25 and retired in 1914 at age 62. The worker’s contributions and accumulating assets were allocated 60% to stocks and 40% to bonds, and the portfolio was rebalanced annually to maintain the allocation. We calculate income on assets using prevailing annual rates of return in the U.S. stock and bond markets during the worker’s career. When the 62-year-old worker retired in 1914, the accumulated retirement savings were converted to an annuity based on the interest rate at year-end 1913. In the second period, the worker started saving in 1878 and retired in 1915 at age 62. The third period was another year later, and so on until the last period, which started in 1976 and ended with the worker retiring in early 2013.

In Figure 3, the teal line shows workers’ allowable accumulations under the proposed lifetime limits. Each worker’s eventual annuity was the result of the age-25 base amount of $50,000 incremented by the annual inflation rate until retirement at age 62. The allowable accumulation was the lump sum equivalent of the cap based on interest rates at the end of the year before retirement. The black line shows what happened to the worker’s projected accumulation under current-law limits on covered compensation and contributions as well as the limit on lifetime accumulations. 

Figure 3. Allowable accumulations under proposed lifetime limits compared with accumulations under contribution, compensation and lifetime income limits, 1912 – 2013
Towers Watson Insider - Figure 3. Allowable accumulations under proposed lifetime limits compared with accumulations under contribution, compensation and lifetime income limits, 1912 – 2013

Source: Developed by the authors as described in the text

In 70% of the simulated periods, the worker’s accumulation at retirement age was less than the proposed lifetime limit would allow, and accumulations never reached the lifetime limit after the late 1960s.

Retirement accumulations under flat-dollar lifetime limits

To simplify implementation, some are proposing a flat-dollar limit on retirement accumulations, such as a cap of $2.5 million or $5.0 million, which would allow a constant-value accumulation across time. If a worker retiring in 2013 could accumulate up to $5.0 million, a counterpart retiring 10 years later — assuming adjustments for 3% inflation per year in the intervening period — could accumulate $6.72 million. 

We consider the tax-preferred retirement accumulations of a highly paid worker who started saving in 1976 and turned 62 in 2013, given historical price inflation, investment returns and interest rates over the period. We also assume current compensation and contribution limits were indexed by annual inflation, and that permissible annual accumulations were discounted by the worker’s remaining years until retirement. As in the analysis of the administration’s proposal, if a worker’s accumulated savings reached the regulatory limit, contributions stopped until increases in the lifetime limit made room for additional contributions. 

Figure 4 shows three simulations for this worker. One simulation applies current-law contribution and compensation limits with no lifetime limitation. The second and third apply these in combination with 2013 lifetime limits of $2.5 million and $5.0 million, respectively, with both adjusted annually by the CPI. All three scenarios assume the worker contributed the maximum allowed each year, and contributions and accumulating balances were invested 60% in stocks and 40% in bonds with annual rebalancing.

In Figure 4, the first simulation (top line) shows potential accumulations for the worker assuming current compensation and contribution maximums were the only constraints. In this scenario, the worker accumulated $2.5 million, almost exactly the amount that would be allowed under the $625,000 lifetime limit cap in 1976 — the indexed equivalent of $2.5 million in 2013. In the second simulation, the worker was subject to a $1.25 million lifetime accumulation limit in 1976, which indexed to $5.0 million in 2013. In the third simulation, the worker’s retirement savings were subject to the $2.5 million cap, which was $625,000 in 1976.

Under the $2.5 million lifetime cap, the worker’s accumulated retirement savings at age 62 were just under $1.6 million — only 64% of the stipulated lifetime cap in 2013. Under the $5.0 million lifetime cap, the worker’s age-62 accumulation was slightly less than $2.1 million — only 83% of the nominal cap at retirement. The lifetime accumulation under the $5.0 million proposal was only 29% larger than the accumulation under the lower lifetime limit.

Figure 4. Worker’s retirement savings from 1976 to 2013 under current contribution and compensation limits and under $5.0 million and $2.5 million caps
Towers Watson Insider - Figure 4. Worker’s retirement savings from 1976 to 2013 under current contribution and compensation limits and under $5.0 million and $2.5 million caps

Source: Developed by the authors as described in the text

Under the $5 million fixed-cap scenario, when the worker was age 25 in 1976, the nominal limit was $1.25 million and the annual contribution limit was $12,000. By the time the worker was 31, high inflation had driven up the lifetime limit to $2.1 million and the annual contribution maximum to $21,000. But the lifetime accumulation for this worker discounted to age 31 was only $27,000 because of the high interest rates in 1982. This worker could not have made any contributions at age 31. If interest rates at the time had been 5% rather than nearly 15%, that $27,000 cap would have been nearly $440,000. This worker lost the opportunity to save early in the career and thus also lost potential earnings on those foregone contributions. These losses could not be made up later through higher contributions because of the annual contribution limits.   

To figure out how all the moving parts would have worked together historically, we assume our hypothetical worker contributed consistently from age 25 onward under initial lifetime limits of either $625,000 or $1.25 million at age 25, and an annual contribution limit of $12,000, all of which were indexed by the CPI. We simulate the worker’s accumulation through serial investment and inflation results for 100 unique historical periods, each equivalent to the duration of our hypothetical worker’s career.

In the first period, the worker started saving at age 25 in 1877, contributions and accumulating assets were allocated 60% to stocks and 40% to bonds at prevailing rates, and the worker retired at age 62 in 1914. The second simulated period started one year later and ended one year later, and so on until the last period, which started in 1976 and ended with the worker’s retirement at the beginning of 2013. As with the earlier case, the maximum allowed in the retirement account was the lifetime limit for that year discounted by current interest rates from age 62 back to the worker’s current age.

Figure 5 shows what happened when the worker was subjected to the cap of $625,000, starting at age 25. The teal line is the nominal value of the indexed cap at retirement. The red line is the accumulated value of the account under existing contribution and compensation limits only. The black line is the accumulated value of the account given the operations of the $2.5 million lifetime limit in combination with current-law compensation and contribution limits.

Figure 5. Effects of lifetime savings cap (equivalent to $2.5 million in 2013) on retirement savings for high earners from age 25 through 61 under different financial market and consumer price histories, 1912 – 2013
Towers Watson Media

Source: Developed by the authors as described in the text

Over virtually all retirement years, accumulated balances under the current compensation and contribution limits were higher than those under the lifetime caps. In the early periods, the lifetime caps reduced retirement savings by up to one-third compared with the compensation and contribution limits only. The differential dropped to less than 10% during the early 1990s but shot up to more than 35% in recent years. From the early 1970s through the late 1980s, the contribution limits constrained accumulations to a greater extent than the lifetime limits. The proposed lifetime limits would have reduced lifetime savings by around $1.0 million for high earners retiring in the last 15 years.

Avoiding unintended consequences

One purpose of lifetime limits would be to keep highly compensated workers from getting more than their fair share of the tax preferences for retirement savings. Such concerns are fueled by occasional media reports of enormous amounts being stashed in IRAs, such as during the 2012 presidential campaign, when disclosures by Republican candidate Mitt Romney suggested that his IRA held between $20.7 and $101.6 million. His total wealth holdings were estimated at between $84.8 million and $264.7 million, and as reports suggested that Romney’s total wealth was toward the upper end of the range,4 the relative importance of the IRA in his total holdings suggests that its value was also toward the upper range.

An IRA accumulation of roughly $100 million indicates that Romney’s retirement plan assets performed considerably better than assets more generally available in the U.S. economy. In The Atlantic, William Cohan explored how Romney could have amassed roughly $100 million in his IRA in just a few short years, and the wealth path was very specific to private equity partners in firms such as Bain Capital.5 Another larger-than-life example would be Bill Gates, who might have been able to accumulate that much by holding personal tax-qualified retirement plan assets in Microsoft stock before its meteoric rise.

If Romney-style IRA holdings were commonplace, it would make sense for regulators to introduce lifetime accumulation limits. But for every Bain Capital or Microsoft, there is an Enron where workers invested all their retirement money in company stock and the result was catastrophic losses rather than outsized gains.

A recent analysis by John Turner, David McCarthy and Norman Stein provides some sense of the scope of the issue.6 Citing an analysis developed by the Employee Benefit Research Institute, they estimated that, of 20.6 million IRAs in 2011, only about 6,180 held balances of more than $3 million. They examined Form 5500 filings for 2010 with the Department of Labor to find plans with large asset balances and small numbers of participants, and identified 10 plans where average account balances ranged from $11.9 million to $253.8 million. Using estimates from the 2010 Survey of Consumer Finances, they estimated that roughly 176,000 households had defined contribution plans plus IRA balances of $3 million or more. Of these, roughly 122,000 households had balances between $3 million and $5 million dollars — amounts that would conceivably remain possible under a $2.5 million lifetime cap if the household included both a husband and wife with balances of that amount. Of the remaining 54,000 households with large balances, 51,000 of them had balances between $5 million and $10 million. 

So the problem of wildly excessive accumulations does not involve colossal numbers of retirement savers. If policymakers aren’t careful, the solution could potentially inflict significant and unanticipated damage on many workers whose retirement savings were not a concern in the first place.

Even allowing that a few outliers might accumulate outsize retirement savings in unusual scenarios, one might question the practical wisdom of holding such enormous balances in a tax-preferred retirement account. Retirement accumulations are taxed as regular income when the assets are paid out to retirees. Whereas after holding such assets in a non-tax-advantaged investment portfolio for more than one year, the initial investment will be considered post-tax income, with the payouts taxed at the lower capital gains rates.

Moreover, under an IRA or other tax-preferred account, the benefits must be distributed when the account holder turns 70½. Given current life expectancies, the distribution would be the account balance divided by 27.4. For a $100 million tax-qualified retirement account whose owner turned 70½ in 2013, the required distribution would have been $3,649,635. The federal income tax on the distribution would be roughly $1,445,255 (at top tax rates), while the same distribution from a non-tax-preferred account would be subject to capital gains taxes of $729,927. 

In a non-tax-preferred account, the original investment would have been made with cash that had already been taxed as income, but where returns were abnormally high, the taxes on the original investment would have been relatively small. The total taxes would likely be much lower than the eventual taxation of investment income paid out from a tax-preferred retirement account. To the extent the investment results come as a surprise, holding the money in a tax-qualified plan could even become an expensive tax trap.

The current compensation and contribution limits are effectively keeping the vast majority of workers from accumulating excessive balances. But if regulators do decide to limit retirement savings in tax-preferred plans and accounts, it will be important not to disadvantage workers whose primary income is their salaries and whose retirement savings are already constrained by the compensation and contribution limits. The overwhelming majority of these are career workers — not mega-rich investors — who will depend on their retirement benefits to see them through retirement.

Young and middle-aged high earners already face much stricter limits on retirement saving than older workers, and introducing a lifetime cap would reduce their retirement savings prospects even further. Given the way such limits are applied over a career, the restrictive effects of the proposed caps would go beyond regulatory intent. It will also be important to moderate the effects on relatively low accumulations in younger workers’ accounts because of occasional high volatility in financial markets. While the $5 million limit might seem reasonable, discounting the $5 million by the interest rates in effect when a worker is 30 or 40 could limit some to much less than the regulatory maximum at retirement age due to interactions with the other limits.

One work-around would be making the limit somewhat higher than $5.0 million — say $7.5 million — but requiring that any excess over $5.0 million be distributed before age 70½. In this fashion, super-investors would be capped much earlier in their careers, those who are generally governed by the compensation and contribution limits would continue to be so governed, and the few who do marginally better than general markets would not have their retirement saving disrupted by anomalous swings in the financial markets.

This article is a summary of a longer article that will appear in the next issue of Benefits Quarterly


1. See Sylvester J. Schieber, The Predictable Surprise: The Unraveling of the U.S. Retirement System (Oxford: Oxford University Press, 2012), for a full discussion of these developments.

2. Robert J. Shiller, data found at: www.irrationalexuberance.com/index.htm.

3. Sylvester J. Schieber, The Predictable Surprise, pp. 166-167.

4. Mark Maremont,Romney’s Unorthodox IRA,” The Wall Street Journal (January 19, 2012) U.S. edition.

5. William Cohan, “What's Really Going on With Mitt Romney's $102 Million IRA,” The Atlantic, September 10, 2012.

6. John A. Turner, David D. McCarthy and Norman P. Stein, “Defined Contribution Plans with Very Large Individual Account Balances,” The Journal of Retirement (Winter 2014), pp. 113-123.