This year’s Wharton Pension Research Council conference theme was about reimagining pensions over the next 40 years. What did the brightest minds in the world come up with?
Traditional pensions should look more like variable annuities. And defined contribution plans should look more like variable annuities too. Is it time to prepare ourselves for a variable annuity revolution?
Not if you look at variable annuity (VA) sales numbers. VA sales peaked at $184 billion in 2007 and have been over 20% below that high water mark over the last two years. Many insurers got out of the business after going a little crazy on guarantees, and the percentage of consumers with a favorable opinion of annuities hovers in the same range as Congress or venereal disease.
This lack of popularity is happening at the same time the largest cohort of new retirees in the history of mankind is looking for something to do with their $6.5 trillion of IRA assets. How about a single investment that lets them take a reasonable amount of investment risk while protecting against outliving assets? Baby boomers should be knocking down doors to shift their assets into a variable annuity.
How are variable annuities not the most popular financial product in IRAs? It should be noted that the $145 billion of VA purchases in 2013 overestimates their popularity as a retirement tool since many are bought by really rich people using them as a tax shelter of last resort. As we know, a VA in an IRA takes away any sheltering bonus so the product needs to hold its own against increasingly cheap mutual funds and ETFs. VAs need to be lean and easy for a consumer to understand, they need to capture a risk premium and be immune to firm-specific default, and they need to provide actual income that a retiree can spend.
This was pretty much what economist William Greenough envisioned when he came up with the VA concept for TIAA-CREF. The big problem with traditional immediate life annuities is that they invest in safe assets that tend to be suboptimal over a long retirement. A single premium immediate annuity (SPIA) will gradually lose purchasing power over time, and the amount you can spend each year will be calculated based on market bond rates the day you buy it. Recent research shows that most retirees should optimally accept quite a bit of investment risk in retirement, and a SPIA forces them to put a big chunk of their portfolio in a bond-like asset.
The perfect annuity would allow retirees to accept a certain amount of investment risk. What does investment risk really mean in retirement? It means that their income could go up or down depending on the market. A variable annuity shouldn’t mean that the pre-annuitization savings is variable—it should mean that the income received after retirement is variable.
A lot of insurance companies want to take the variability away from a variable annuity by providing stable income guarantees and a springing option of potentially higher income in the future. When any entity guarantees an income for retirees, this creates an important problem for the institution (which may be an insurance company, corporation or government agency). If markets don’t cooperate, then the institution is on the hook to make payments and must scrounge for money from other revenue sources to make up the difference.
This is the transfer of investment risk from the annuitant to the institution. It can make sense, but this transfer is costly (for an institution to accept that risk, it should mean a lower annuity payment on average) and it puts the institution at risk of going out of business and not paying any annuitants.
In a fascinating and important presentation at the Wharton conference, pension expert David S. Blitzstein (who negotiates pensions for the United Food & Commercial Workers International Union) argued for a mechanism to allow greater risk sharing in pensions by providing variable incomes tied to market returns. Let me just say that one more time—a union pension expert argued that it is in the best interest of employees to encourage employers to transfer investment risk from their own balance sheet to union employees.
Why? Because it’s too, well, risky for employers to be forced to dig into their own assets to make up for guaranteed income promises when markets fall (or stagnate). This could mean that no employees get pensions if employers go bankrupt, and you’d rather that employees deal with slightly varying income than no income at all.
This also helps explain why insurance companies shouldn’t necessarily be the ones bearing the investment risk from income guarantees. The other problem is that, in the case of a pension or a mutual insurance company, the other participants lose out when promised guarantees are overly generous—creating a wealth transfer from one group to another.
In the worst case, an insurance company could go belly up in a market crisis. The risk of insurer default can be impossible for a retiree to hedge outside of state guarantee associations that provide as little as $100,000 of protection and may not be rock solid if multiple firms fail. Risk sharing is probably the most efficient way to deal with unknown future investment returns.
Why are variable annuities the optimal retirement income tool? In theory (with giant air quotes), variable annuities are all about giving retirees two very important things they don’t get with traditional fund investments. Unlike withdrawing money from a mutual fund, a variable annuity will allow a retiree to spend more each year from pooling longevity risk and will provide protection against running out of money.
In the absence of a variable annuity, a retiree must figure out how much to withdraw each year from an investment account. This requires some difficult calculus that even the experts can’t agree on. Take out too much or live too long and you’ll run out of money. Take out too little and you’ll have a lot less fun in retirement. Instead of retaining the risk of not knowing how long you’re going to live, it is far more efficient to pool this risk. Keep the investment risk that balanced mutual funds provide, but pool the longevity risk.
A variable income annuity linked to market performance promises both. In many ways, variable annuities after retirement are like lifecycle funds before retirement. They are an optimal retirement investment instrument. Sure, employees could rebalance their own portfolios regularly and select the right mix of stocks and bonds, but we know that nobody actually does that (even financial planning professors).
The most popular investment in the Thrift Savings Plan, largely considered the most efficient DC plan, is a T-bill-based fund that’s paid about 2% over the last five years. Making it easy for employees to default into lifecycle funds gave millions of workers access to a more appropriate investment portfolio and a chance to share in the risk premium to meet retirement goals.
Variable annuities could do the same thing for retirees. Opting a part of retiree assets into a VA would “theoretically” improve their average investment portfolio performance while protecting against the risk of running out of money. In a perfect world, employers could add a VA to a DC plan without fear of being sued by employees for a fiduciary breach. Greater availability of VA products could enhance industry competition and reduce fees. Sounds like a terrific idea. In “theory.”
At this point, if you’re an advocate of existing VA products and have high blood pressure you may want to stop reading. Because the problem with VAs is that the products often don’t look like an optimal retirement income tool. Most of them really look like one of two things—a tax sheltering device for the rich, and a complex financial instrument that resembles a structured product with features that make it more attractive in a sales pitch.
For those interested in a tax sheltering device, the annuity part may actually be irrelevant. As a reminder, investments in a variable annuity are tax sheltered before the so-called distribution phase. After age 59 ½, you can withdraw the money any way you like and pay income tax on the prorated amount that exceeds the initial investment. Withdrawal options can include a lump sum, a fixed annuity payment or a variable annuity payment. To an economist, the efficient withdrawal strategy will be a variable annuity payment. According to a recent study by Ruark Consulting, fewer than 5% of VAs are annuitized annually.
That low rates of VA annuitization are seen as a good thing by the industry illustrates what’s wrong with the VA market today. One of the most telling findings from the Ruark study is that even when annuitants transfer investment risk to an annuity company through an income guarantee, they very often don’t actually cash in on their in-the-money annuity. Again, this is seen as good news for the industry for obvious reasons. But consumer misunderstanding of annuity features is a big problem.
According to FINRA, VAs are among the most frequent sources of consumer complaints for questionable sales practices and high surrender charges. The more complex VA features generally allow a salesperson to claim upside potential with downside protection. Many of these products are sold to less sophisticated consumers and include “gotcha” provisions that lock buyers into contracts or provide inflated claims of investment returns that stretch reality.
A recent paper published in The Journal of Retirement by American College professor Wade Pfau studied the value of these guarantees. Guaranteed living withdrawal benefits (GLWBs) provide a steady income, similar to a SPIA, but include an option that will provide a higher permanent income if a hypothetical benefit base tied to market investments is able to rise beyond its past high water mark. As Moshe Milevsky has pointed out in this publication and elsewhere, the devil is in the details when it comes to how the benefit base is affected by fees and limitations on investments.
Pfau uses assumptions from a no-load, low-cost VA with GLWB rider. Since investors who hold a VA with a guaranteed floor income can take more investment risk, Pfau attempts to compare variable annuity portfolios to less conservative investment portfolios with no VA. He finds that the downside protection of a GLWB is actually lower than a less conservative investment portfolio in real terms.
This makes some sense because the low guaranteed income loses spending power each year and over time becomes a weaker income floor. He finds that by the 30th year of retirement, the step-up feature provides a constant inflation-adjusted income 12% of the time. And over two-thirds of VA/GLWBs have depleted their contract value after 30 years.
The upside is also greater because investors can take more risk, but there is a problem. If you hold an investment portfolio outside the variable annuity, then you really need the VA to perform well when your investments are falling. But the VA only converts to a higher income when investments are doing very well. The low risk-hedging value of existing guarantees is one reason Milevsky has proposed a deferred income annuity product that triggers only when investment returns do poorly enough to place investors at risk of depleting retirement savings.
Pfau provides a fascinating table showing the probability of a step-up falls rapidly by year of retirement to just 10% by the 10th year. If the variable income option is going to spring on a VA/GLWB, it’s probably only going to happen in the first five years before fees and withdrawals eat too far into the contract value.
In a recent seminar at Texas Tech, Manish Malhotra, developer of the Income Discovery retirement software simulation, estimated the impact of VA/GLWBs on the likelihood of income shortfall later in life and found that they are most beneficial when markets do well and the income base grows. But in most simulations, when investors need it the most the VA looks like a SPIA with a lower payout. Generally, this reduces the likelihood that a retiree will meet his or her annual spending goal.
A product with a relatively low nominal income base and a possibility for modest increases in the first few years of retirement doesn’t really sound like the best way to accept market risk and provide inflation protection later in retirement. But it does provide a good story that can be used to convince investors who fear both low investment returns and stock market risk to buy a variable annuity.
VAs can either be sold or they can be bought. The mutual fund market has been transformed by the shift from higher-commission products with higher fees that must be sold to more streamlined products bought by more sophisticated investors in taxable accounts and, most importantly, selected by workers in retirement accounts. But these workers will need to do something with all that mutual fund money when they retire. It makes sense that they have a range of equally streamlined annuitization options.
VAs, of course, have a product problem. First, unless you’re Moshe Milevsky or Wade Pfau, it can be nearly impossible to understand what you’re buying. The range of product features coupled with opaque pricing means that convincing a plan sponsor to include them in a DC plan is an uphill battle. At least mutual funds have an easily understood way to compare expense ratios.
At a recent conference, I suggested the industry get together and decide on two or three VA product structures that made the most sense for an average retiree. Then provide standardized cost disclosure and let companies compete to provide these products within qualified plans. Standardization and competition should make plan sponsors more comfortable including them. This way, VAs can more easily be bought and not sold.
The idea, of course, was a nonstarter to companies that prefer the consistent, higher margin income from existing products. But turning retirement assets into income is an important social problem that the industry and government need to work together to solve, and VAs, like target date mutual funds, are the right way to solve that problem.
Are VAs going to join the self-directed retirement revolution in America? Does it make sense for the industry to starting seeing VAs through a DC lens? Target date funds attracted $50 billion in new fund flows in 2007. In 2013, that was over $500 billion. There’s an open door at the decumulation party. Stepping through it might mean leaving some outdated product models at the door.