Mandating annuities in retirement

By Felix Salmon
September 29, 2010
Dan Ariely had an interesting column in the latest issue of HBR, talking about how Chile forces its citizens to save money and annuitize their pensions:

" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

Dan Ariely had an interesting column in the latest issue of HBR, talking about how Chile forces its citizens to save money and annuitize their pensions:

When employees reach retirement, their savings are converted into annuities. The government auctions off the rights to annuitize retirees in groups of 250,000…

Institutionally, Chile has cracked an age-old problem with annuities. It’s risky business to predict how long people will live, so insurance companies charge a high premium to cover that risk, which makes for an inefficient market. Annuities also suffer from an adverse selection problem… By pooling the risk, the Chilean government makes annuities an attractive business with more competition and better prices. And since everyone is forced to annuitize, the adverse selection problem simply disappears.

This is rather clever, if it’s true. But a Chilean technocrat, Axel Christensen, responded on the HBR website, saying that Ariely misunderstood what he’d been told. The groups of 250,000 are allocated to fund managers, he said, not annuity providers.

At retirement, Chileans may choose between a fixed inflation-adjusted annuity offered by an insurance company or a variable annuity from by the same company that managed their retirement account. It is an individual decision, with no pooling as you stated. The insurance companies have to bid for the contributor´s savings that increases competition, but the system does has its flaws, like the adverse selection you identified.

This doesn’t clear things up a lot: it seems to me that if you have to pick an annuity, then adverse-selection problems are minimized even if there’s no pooling at all. After all, the problem with adverse selection is that people who buy annuities will live longer than people who don’t buy annuities. If everybody buys an annuity, there isn’t a problem.

And when Ariely republished the column on his blog after Christensen had made his comment, the column was unchanged. I don’t know what to make of that: maybe Ariely didn’t see the comment, or he thinks that for some reason it’s unimportant.

Ariely says that schemes like Chile’s wouldn’t go down well in the U.S., where Americans would consider it “heavy-handed and limiting”. I daresay he’s right. But it would be great if there were some way of allowing people to voluntarily commit to annuitizing their pension fund upon retirement. One of the problems with pension funds is that nobody actually needs some big multi-million-dollar nest egg at age 65. What they need, instead, is a healthy income in retirement. But converting a nest egg into an income is non-trivial. You want to maximize your income by spending principal as well as interest, but you also want to make sure you don’t run out of money if you live a long time.

Annuities solve that problem, but they do suffer from adverse selection: people who buy them live longer than people who don’t, and so insurance companies have to make allowances for that. If everybody in a big pool was committed to annuitizing, then the insurance company could ensure that people who died at a younger age would help to subsidize those who live a very long time — as should happen in any good pension system.

This, indeed, is one of the central problems with defined-contribution pensions rather than defined-benefit pensions. When we “save up for retirement”, we’re conflating two things: the savings, on the one hand, and our retirement income, on the other. If we die before we retire, then our retirement income is zero, but the savings are still there, and the only retiree they’re likely to benefit is our spouse, if we have one.

Are there any numbers on the amount of money which is paid in to Social Security against which no benefits are ever drawn? I’d include people working in the U.S. on temporary work visas, here, as well as people who die before retirement while unmarried. On top of that, of course, people who die early in retirement end up taking out of the system much less than they put in. And the benefits of that cross-subsidy accrue to the long-lived, who need money to live on in their 90s and beyond. It’s a humane and sensible system: the living need money more than the dead do.

Off the top of my head, I can’t think of a way of replicating anything like this on a voluntary basis. I could invest my retirement savings in a fund which automatically annuitizes with everybody else in the fund when I turn 65, for instance. But if I get cancer when I’m 64, I’ll surely move those savings into cash instead of meekly accepting a short-lived income.

So the Chilean system of mandating annuitization for certain types of retirement assets makes sense to me, if indeed there is a mandate there. Maybe people could have a choice: they can invest pre-tax dollars in retirement funds which are forced to annuitize, but if they want to retain control over whether or not they annuitize, they have to invest only post-tax dollars.

And then, of course, we’d have to look to see whether the insurance companies actually improved their annuity rates significantly in response to the new mandate. Is there any data from Chile on that? All of this government interference might make sense in theory, but the real world is nearly always much messier.


We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see

Last time I checked (2/3 years) Chileans had 2 choices, buy an annuity or go for “programmed withdrawal” a scheme similar to the british “income withdrawal” that’s not an annuity.

Posted by alea | Report as abusive

Sorry, Felix, but from my perspective this is a TERRIBLE idea. We expect to need about $40k of income (adjusted for inflation) in retirement. No mortgage, kids five years out of college, a modest middle-class lifestyle. As the discussion of “Schwarzenegger’s pension math” proved, that is the rough equivalent of a million dollars in retirement savings. (Assuming the Social Security has been put out of its misery by then.)

Yet that is a grossly insufficient accumulation! Nursing-home or in-house assistance can rapidly get very expensive. And Medicare, by then, will be operated on a cost-conscious basis. Suppose I place a higher value on my life than the Medicare boards? This portion of savings should NOT be annuitized. If it is never needed (and most people eventually need long-term care), we can ultimately pass it along to our children for their personal security.

Please let me manage my own investments and plan my own retirement. I don’t trust mutual fund managers to have my best interests in mind. I don’t trust Medicare managers to have my best interests in mind. And I absolutely don’t trust the government to choose wisely.

Posted by TFF | Report as abusive

I’m struggling to get Felix’s point, or maybe don’t understand his fundamental problem is with voluntary annuities.

Personal savings (some tax-deferred, some not) are not the same as a pension system – why must they solve the exact same goals?

“Ensure that people who died at a younger age would help to subsidize those who live a very long time – as should happen in any good pension system.” That is NOT the point of personal savings, tax-deferred or not. It IS the point of defined benefits plans and Social Security.

“This, indeed, is one of the central problems with defined-contribution pensions rather than defined-benefit pensions.” No, it is one of the central DIFFERENCES between the two. They are different constructs with different social goals. Felix wants to treat them as the same thing, and sets about ways to turn the former into the latter. if that’s the goal, then just mandate defined-benefit plans.

“But it would be great if there were some way of allowing people to voluntarily commit to annuitizing their pension fund upon retirement….Annuities solve that problem”

So is your problem that more people don’t voluntarily buy annuities with their savings, and that the best way to solve the inherent adverse selection in a voluntarily-purchased product is….to make it non-voluntary? That IS getting a bit heavy-handed.

Just say you prefer mandated defined-benefit plans over voluntary defined-contribution plans, and cut to the chase.

Posted by SteveHamlin | Report as abusive

I didn’t think annuities have a significant adverse selection problem. That’s why insurers love them. Unlike life and health insurance, unhealthy people don’t buy them. Healthy people do, but they are in no better position than the insurer to predict their own mortality. It’s purely actuarial. It’s sort of the opposite of the lemons problem that plagues life and health insurance — with annuities, there is no significant information asymmetry problem, which allows the product to be priced correctly and efficiently.

Posted by maynardGkeynes | Report as abusive

I’m pretty sure we already have this system. It’s called Social Security.

Posted by MitchW | Report as abusive

And the overhead taken off the top, the fees etc., as contrasted to the overhead of social security, ? So our savings would be locked into a system where they go into the Wall Street Casino? I shudder to think of handing my savings over to the experts who have done so well the past few years.

Posted by Laumilo | Report as abusive

annuities should be an option, as far as I can tell that is all the new rules require. its not mandatory. and i too am unsure how there is a problem with them for insurance companies. they don’t have to worry about paying out more for some than others, since any body can buy them. as for why savings and pensions/SSN are the same? they are trying to address the same problem. retirement. problem is one has professionals doing investment full time, the other doesn’t. but they both depend on equities and bonds to make the retiree’s plan works. so far, 401/ira/savings have one other problem, they don’t know how much is needed to save. thats because we can’t tell the future of how long some one will live, and how much it will cost to live that long. don’t know the answer to those 2 questions and you have no idea how much is needed to be saved. we do know that health care costs have been rising for decades even in the midst of a recession. and we know insurance companies won’t touch those on Medicare for any amount of money (they know its instant claim).

Posted by willid3 | Report as abusive

Ummm…..The Chilean system is consistently overhyped.

The fees and total expense ratio are high by commercial standards, and horrific compared to social security, eating away at the principal, and as age cohorts start selling equities, they depress prices when they need the cash, so it’s a recipe for buy high sell low.

When people start talking about privatizing social security, and they start talking about the Chile success story, I know that they are fools.

Posted by Matthew_Saroff | Report as abusive

We already know how this story plays out… The investments offered by 401k plan have expense ratios that average above 2%, as compared with under .2% for low-cos index funds.

This a classic agency problem. Employers (or government selected annuity providers) can’t be expected to watch out for the interests of workers as well as workers would watch out for their own interests.

Immediate life annuities can be a good choice for retirees, but only if they’re offered in a competitive market, where the consumer gets to choose.

As an aside, 401k plans should be changed immediately to allow any plan participant to roll their entire balance to an IRA at any time, without having to quit their job. This would force 401k plan providers to compete again and would end the days of plan providers wining and dining financially illiterate HR folks and scoring multi-million dollar lard-fests at the expense of hapless employees.

Posted by Magellan1234 | Report as abusive

@ Magellan — The government TSP plan (like a a (401k)) is extremely well run, with extremely low expenses, run absolutely on behalf of the interests of the employees. That is because it gets close scrutiny from the Federal Government, which private 401K plans do not.

Posted by maynardGkeynes | Report as abusive

The Swiss run a compulsory private system with 100% participation. Employees and Employers must pay 5% each of pensionable salary into employees’ pension pots which must grow by at least 2.5% a year until retirement; after that, a government set annuity rate, higher than the free market rate, is paid. It seems to work rather well. The high participation rates ensure good economies of scale, even in such a small country.

Posted by FifthDecade | Report as abusive

The UK system requires you to buy an annuity. To avoid the gaming you suggest, you cannot withdraw your retirement savings from your pension fund (above a small amount) at any point so all those with retirement savings spread the risk among each other. It’s a basic insurance principle – the risk of the long lived is shared by the short lived. The quid pro quo for this constraint is the tax relief that you get on pension saving as opposed to any other kinds of saving.

To understand this setup you have to understand what the government cares about. They do not care about your lifestyle in retirement or whether you buy a yacht or choose to pass money onto your children or anything. All they care about is that you do not end up reliant on social security – so in return for the tax break you must buy an annuity.

Although the debate in the UK is about freeing this up (and the government is set to loosen the restrictions on buying an annuity) academic studies suggest that for many, buying an annuity is absolutely the rational choice and studies have questioned why more Americans who have that choice don’t make the choice, voluntarily, to insure themselves against living too long.

Posted by Mostannoyed | Report as abusive

“…people who died at a younger age would help to subsidize those who live a very long time — as should happen in any good pension system.”

Really? Is that a given? People who die early not only DIE early, but they also get taxed for it?

I can see where you’re coming from, but if someone has made a personal choice to smoke, drink, rock and roll with the intention of saving little and dying early, shouldn’t they be allowed to?

Posted by LeighCaldwell | Report as abusive

Mostannoyed, I’d be interested in seeing the assumptions in those academic studies that suggest buying an annuity is the rational choice.

If you don’t have enough saved, your best bet may be to use it up faster than an annuity would allow, and then (if you happen to be unlucky enough to live to a ripe old age) throw yourself at the mercy of the government.

If you have more than enough saved, then there is little risk of outliving your savings and thus no strong reason to pay insurance companies their cut on the product.

I’m not saying the studies are wrong, but I’d want to double-check the details before accepting them.

Posted by TFF | Report as abusive

Leigh Caldwell – you’re forgetting this is insurance and this is how insurance works. It’s not – dread word to Americans – socialism! It’s just insurance which is based on pooling of risks. There’s no moral weight attached to the outcomes. You insure your house and you pay every year and if your house goes up in flames you get a payout made up of all the premiums paid in by all the other house-owners who didn’t have a fire. Same with longevity. You have the (in retirement terms) bad luck to outlive your savings – you get a payout (meaning an ongoing annuity payment) made up of all the premiums paid in by people who didn’t live as long.

TFF, it’s been a long time since I worked in this area but if I can dig out the studies, I will. I think the studies actually suggest that people live longer than they think they will – and that’s why it’s rational for some (and it’s only for some) to buy an annuity. But people find it very difficult to predict their own longevity (despite their better knowledge of family history). Hence the tendency not to buy an annuity given a free choice in the matter. I suppose many would also choose not to buy car insurance if they had the choice. People think they are more in control of their own safety in a car than they really are.

Posted by Mostannoyed | Report as abusive 0Short%20Report.pdf

Some interesting stats in here suggesting individuals underestimate average longevity and their own longevity in comparison to the average. At the end, where individual underestimates are given (from a British study) women underesimated their lifespan by almost 6 years and the men by 4.5 years.

Moral of the story is, people think they will be able to manage this risk without insurance when, from a purely actuarial standpoint, they may just be better off using insurance to assist them manage this risk.

Now, if they decide they want to live with the risk for the possible benefit of, say, leaving money to the children in the event of their early demise, then that is (for Americans at least) up to them. But the studies suggest that most people are not good at understanding and measuring the risk and therefore may not be making the most rational decision.

Posted by Mostannoyed | Report as abusive

Ah, I see… I agree at least that far. Estimating the longevity of an individual is almost impossible, which is why (if self-insuring) you almost have to plan on living at least 30 more years into retirement.

I guess my problem is that the numbers I’ve seen seem to have a very large profit margin built in (either that or a very low ROI). At a 4% annual real return, a 30-year inflation-indexed annuity of $3000/month has a present value of $628k. Extending the life expectation has minimal effect. The annuity could last 100 years and still only rise to a PV of $880k. Yet the Schwarzenegger article (and subsequent discussion) priced a lifetime inflation-indexed annuity of $3000/month at $1 million!

If my life expectancy at retirement is 30 years, and I can manage my investments for a 4% real return, then the insurance company is overcharging by a factor of 50%. When the profit-premium is that high they are no longer insuring against longevity — they are insuring against the possibility that your investments have a real return that is less than 4%.

So what real return is necessary to justify a $1M price tag on a 30-year inflation-adjusted annuity? Seems the insurance company is willing to pay me a piddling 0.5% real return on my money! Thank-you-very-much, I’ll do it myself.

Posted by TFF | Report as abusive

As an aside, most of the annuities sold seem to be fixed — not adjusted for inflation. You get a much higher initial payout on a fixed annuity, but the longevity risk attached to such a product is very scary. I don’t want to be 10 years into retirement and living on half the purchasing power that I bargained for.

Posted by TFF | Report as abusive