Christine Benz: Hi, I'm Christine Benz for Morningstar.com. I recently attended the Morningstar Ibbotson Conference and had the chance to sit down with John Ameriks, who focuses on retirement research for Vanguard. We discussed the role of annuities and why many retirees tend to be resistant to buying them.
John, thank you so much for being here.
John Ameriks: I'm happy to be here, Christine. Good to see you.
Benz: You are going to be talking today about the role of annuities in retiree portfolios, and there has been a lot of academic research pointing to annuities being additive to retirees' long-term plans. Let's talk about that data and why so many academics tend to be positive on the role of, say, a single premium immediate annuity?
Ameriks: Sure. I mean, I'd be happy to do that. We'll put it in the context of the broader problem that retirees are trying to solve when they retire. In most academic models, so to get to the theory thing right upfront, in most academic models, the baseline assumption is someone who is saving cares about their future spending. That's why they save because they want to spend at some point in the future, and they are forward-looking when they're doing that.
If you make those assumptions and a bunch of other mathematical assumptions, essentially you will get a pattern of wealth accumulation that looks like a triangle. People build wealth up during the accumulation phase, and they should be spending it down during the decumulation phase.
In most of those models there is no scope or room for the value of assets that are left behind right? If someone dies prematurely and hasn't spent assets, in a sense, they are wasted. It produces no value for them. That's a very important aspect of why annuities in such a framework add so much value. Because what annuities do essentially that insurance product is a pooling arrangement, where a group of people get together, put all their money in a common investment and basically will say, "Look, those of us that survive, will continue to take the average amount of supportable payments out of the pool; while those of us that die will not make a claim on the residual assets in the pool."
You can kind of see immediately if I talk about it like that, that leaves assets in the pool that can be paid out to people who live longer. So, in a sense, if all you care about is the maximum amount of spending over your lifetime, an annuity can add a lot of value because it shifts assets from a state that you don't care about them in, because you're gone, to a state in which you do. And that is why in most models, when you run math, if you got a risk-averse individual, an annuity can add a lot of value. There are lots of other assumptions, though, in the real world that push back on that very basic framework, and I'd love to talk to you about those, too.
Benz: Yes, let's discuss that because when you look at the data and people are confronted with this real world option of "You can either take an annuity payment from, say, a pension, or we'll give you the lump sum," most people choose the lump sum. So, there seems to be this impediment to annuitizing, and your work has taken a closer look at why that might be. Let's talk about some of those factors.
Ameriks: Sure. So, I probably would change the words a little and say rather than an impediment--because as far as I can tell, there really aren't lots of impediments to annuitization--there are a lots of disincentives. There are a lots of reasons that people will give you why they're not interested. So, I've talked about this before; it's sort of death by 1,000 cuts because while in the framework that I articulated, annuities are so great, once you start looking at several aspects of the real world, they lose their attractiveness.
One of those big ones is costs. And in a lot of annuity arrangements, in particular, annuity arrangements available on the private market, insurers have to worry about what's called adverse selection. That is simply the fact that people who look and seek out insurance tend to be ones who are exposed most to the particular risk that they are trying to insure.
Benz: Healthier people who will live a very long time.
Ameriks: That's exactly the same reason why when anybody gets a life insurance, you generally are going to have a medical test because an insurer obviously doesn't want somebody on death's door signing up for a life insurance policy.
In the same way, the insurer doesn't really want, or wants to try to control the number of people who are very, very healthy and are going to live a long time, so that [the insurer is] not gamed out of that. That creates a drag and it creates a price increase over what could be charged if the entire population were to purchase what's call an actuarially fair annuity, which reflected average mortality. The problem is the mortality of annuitants is not average mortality. So, that's why there is a difference.
Benz: It tends to be higher. So there is that adverse selection.
Ameriks: Lower mortality, higher longevity, yes.
Benz: And then there are some other considerations that people might have, things that they might want for their lives or things they might be concerned about. Let's talk about some of those.
Ameriks: Sure. So one of the other very important aspects of a lot of the theoretical work, in particular, the very early work that was done on annuities is, there is an assumption embedded in all of that that annuity arrangements are reversible. That basically while I can be in an annuity pool, I can go back at any moment and get an actuarially fair cash equivalent of the value of my annuity back as a lump sum. In the real world that is next impossible to do, certainly not without very large fees and in some cases literally, it's not legally possible to do that.
So, the irreversibility aspect of annuitization can present a very large disincentive for people to go that route because not all of the expenses that they have will come at them $50 a month or $100 a month. Many things that they're worried about come in much larger chunks and being confined to just taking periodic payments can make it very difficult to meet those expenses. And so people worry about that.
Benz: So, there is a mismatch essentially. You're getting the steady stream of payments.
Ameriks: That's right.
Benz: And your work points to one of the key things that people are concerned about are very high health-care costs later in their lives, long-term care costs, specifically.
Ameriks: Right. So, there are two very important reasons why people hold on to assets into retirement as opposed to exchanging those assets for a promise of a periodic income for the rest of their lives. One is precautionary motives that include this issue around health. I could get very sick. I may have some insurance coverage, but maybe it's not perfect or maybe I'm worried about things that I will want that aren't necessarily covered by insurance, so that's a big part of it.
The other one that's important is, going way back to the beginning of our conversation, this assumption that assets don't matter in the state of the world in which I'm not around. There are a lot of people who see at least a residual value for the wealth that they've accumulated. While it's probably not the optimal use of their funds to have it left behind and not get to spend it, if it goes to say their children or a cause they care about, that is not a zero-value event to them ex-ante. They think about that and it kind of makes them feel good that, "Well, if I don't need the money, my kids or something I care about will get it."
Once you introduce those two motives--precautionary motives and a bequest motive--the motive for annuitization, particularly the result that you would always annuitize everything in retirement, goes away to a large extent depending on the strength of those motives. Annuities are still very valuable, and it gets very complicated.
So, my academic colleagues that I've been collaborating with--Andrew Caplin at New York University, Stijn Van Nieuwerburgh, Steve Laufer, and others--and I put together a model that looks at this and tries to come up with an empirical framework to identify the strength of these two motives--the precautionary motive and bequest motive--and then measure based on the strength of those two things how much annuitization would we expect. And the basic result there is the strength of those motives varies wildly across the population. So, they are relevant to some, irrelevant to others, and the notion that one-size-fits-all is really going to be problematic in retirement income.
Benz: So, there are a few alternatives that at least partly address some of the issues with annuities. One is a product called longevity insurance, essentially an annuity that kicks in only if you exceed your life expectancy, if you live longer than that. Let's talk about that product and its viability.
Ameriks: One of the things that we explicitly looked at was an annuity contract that worked this way. And when you put that in the context of this model that we've built, that gives you a sense of what annuity demand would look like. Because of that type of product design, which involves income only if someone gets to a very advanced age, that doesn't require a large lump sum upfront. It's a much smaller commitment for someone to make, particularly in intervening years when a lot of these precautionary events may occur.
That increases demand in the models for that kind of a solution. The issue there is you always question this. You can build a model and say, "Look, there should be more demand for this." Obviously, this product that we're talking about, longevity insurance, it does exist in the marketplace, but it's still not sold very heavily, even less than broad single-premium immediate annuities.
In the real world, there are still issues with other things that I didn't mention. I mentioned a 1,000 cuts. We only talked about a couple. There are more, particularly coming out of the financial crisis. I think issues around the solvency of insurers and their ability to make very, very long-term promises is something that weighs on retirees minds. There are tax considerations that up until recently--I will undo what I said a little earlier about impediments--the tax laws around how longevity insurance was going to be treated with respect to required minimum distributions have introduced some uncertainty there, if not an impediment, and that looks squared to go away.
So, I can say that it is something that I think we're actively looking at and interested in as another tool to put in the kit that people can use to attempt to build what works for them in retirement.
Benz: John, it's obviously a very complicated topic, lots of different angles to look at it, but we appreciate you being here and sharing your insights. It's always helpful.
Ameriks: Anytime, Christine. Thank you.