Evaluating SPIAs by implied return or life expectancy

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
petulant
Posts: 3601
Joined: Thu Sep 22, 2016 1:09 pm

Evaluating SPIAs by implied return or life expectancy

Post by petulant »

A Single Premium Immediate Annuity (SPIA) makes periodic payments for a guaranteed time period. While they can be structured for a number of time periods, the most popular form to discuss is the receipt of payments for the remainder of the purchaser's life. BH sometimes discuss SPIAs as one tool used to transform a portfolio of assets into safe retirement income.

Evaluating SPIAs can be difficult because a SPIA quote is stated as a payout rate that looks attractive if compared to the yields or expected returns on other financial products, but this payout rate is not equivalent to the yields on other financial products. A SPIA's payout rate includes the following components:
  • Investment returns
  • Return of principal
  • Mortality credits
Hence, what looks like a 9% payout rate is emphatically not a 9% return on investment. Consider a 15-year mortgage with a 4% interest rate. For each $100,000 in principal, that mortgage would have payments of almost $9,000 per year, or 9% of the starting balance. The actual cost of that debt is 4%, but the payments are 9%. That is what a SPIA with lifetime payments does: it is like a mortgage being paid by the insurance company to the retiree with a lower yield figure plus return of principal.

That means that every SPIA has an implied investment return figure, a period over which principal is returned, and a mortality/actuarial component.

Now, if we knew the time period of payments, the payment amount, and the starting balance, we could figure out the investment return implicit in the SPIA quote. But you don't know how long you live, and the insurance company is using an actuarial estimate. Or, given the payment amount, starting balance, and an interest rate, we could estimate the insurance company's estimate of your life expectancy. But we can't derive a fair life expectancy figure (the payment period) AND interest rate just from a SPIA payout rate.

Hence, the way to evaluate a SPIA payout rate is to get some independent estimate of one figure (the interest rate or life expectancy) and then solve for the other figure, then do vice versa. I'll work out an example.

I begin with a SPIA quote from immediateannuities.com for a 70-year-old female in Iowa paying $100,000. With no cash refund or period certain, the payout is $657 per month, or an annual payout rate of 7.884%.

Estimating Return Given Life Expectancy
To estimate the implied return, I could use an estimated life expectancy like the Social Security Administration's actuarial life table. Per that table, a 70-year-old female has a life expectancy of 15.82 years. So I can plug in the following Excel formula to get a close estimate of the implied investment return in the SPIA: =RATE(15.82,7884,-100000,0).

EDIT: The first term in the formula above is the life expectancy (nper), the second is an annualized measure of the payment (pmt), the third term is the contract premium (pv, a negative number), and the last number is zero for the amount remaining when the annuitant passes away (fv).

In this case, the rate is 2.76%. Now, there are many people who believe the SSA actuarial table is not useful for this purpose because it covers the entire population, while those who purchase SPIAs tend to be wealthier, healthier, and longer-lived. So, I consulted the Bankrate Life Expectancy Calculator, answering various questions for my hypothetical 70-year-old woman in a reasonably healthy but not active way, resulting in a life expectancy of 18.5 more years. Using that for the formula, =RATE(18.5,7884,-100000,0), results in an implied return of 4.2%. So you can see health and life expectancy matter for evaluating the rate.

Estimating Life Expectancy Using Interest Rate Proxy
To gauge the insurance company's life expectancy estimate, we can use a proxy interest rate like the 20-year Treasury bond, which is currently yielding 4.06% (at the time of writing). We plug that into the following Excel formula: =NPER(0.0406,7884,-100000,0).

EDIT: The first term in the formula is the proxy interest rate entered as a decimal (rate), the second are the annualized payments (pmt), the third is the contract premium (pv, a negative number), and the last number is zero for the amount remaining when the annuitant passes away (fv).

The result is 18.18 years. So, this is saying that if the insurance company starts with a rate of 4.06%, its quote is assuming a life expectancy of 18.18 years. As a purchaser, the hypothetical 70-year-old female in Iowa could evaluate whether she believes this figure is reasonable.

Estimating Return with COLA Given Life Expectancy
EDIT: Now, another wrinkle is that many BH would consider getting a SPIA with a COLA. We have a very good approximation to the interest rate estimate above by adding a simple multiple. In this case, I have gotten a quote for a 70-year-old woman in Iowa putting in $100,000 but also getting 2% annual increases. The monthly payments for her would start at $554, or $6648 annually. The formula would be =(1+RATE(18.5,6648,-100000,0))*(1+.02).

The first value 18.5 is the life expectancy (nper), the second value 6648 is the starting annualized payment (pmt), the third value -100000 is the contract premium (pv, a negative number), and the fourth value is the remaining amount after the annuitant passes away, or zero (fv). Note also that the specific COLA is added at the end as .02.

Unsurprisingly, this formula shows a discount rate or implied rate of return of 4.26%, pretty close to the quotes on the non-COLA annuity.

Kudos to #Cruncher for suggesting this approximation.
Last edited by petulant on Fri Jun 02, 2023 1:36 pm, edited 4 times in total.
User avatar
JoMoney
Posts: 16260
Joined: Tue Jul 23, 2013 5:31 am

Re: Evaluating SPIAs by implied return or life expectancy

Post by JoMoney »

I use an amortization formula/calculator based on current interest rates available on a portfolio of bonds laddered out over some reasonable life expectancy. With that interest rate, and the life expectancy as the "loan term" for the amortization function, I plug those in with the amount I'd be willing to invest in a bond ladder or SPIA premium, and if the SPIA is cheaper than the bond ladder it looks like a preferable option... and I add extra weight to one or the other depending on whether I have some preference for the insurance guarantee of the SPIA if I live longer than expectations, or extra weight to the bond ladder preference if I'd rather have a beneficiary get remains if I live less than expectations.

I'll add in years past while interest rates were low, the payoff of a SPIA vs. withdrawing from a bond ladder over amortized over life expectancy was giving a much larger advantage to the SPIA... in more recent times that I've looked, having a bond ladder at current rates present a more competitive option, especially if one is less concerned about living longer than expectancy or have other resources where that's not as a big a factor in the comparison.
Last edited by JoMoney on Wed May 31, 2023 8:27 am, edited 1 time in total.
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham
Chuck
Posts: 2505
Joined: Thu May 21, 2009 12:19 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by Chuck »

petulant wrote: Wed May 31, 2023 7:49 am Now, another wrinkle is that many BH would consider getting a SPIA with a COLA. The simple Excel formulas don't work with a COLA; as far as I know, it would require building a larger spreadsheet and applying discount rates across cells.
Just run the same formulas using real dollars and a real interest rate, and interpret the numbers as such.
dcabler
Posts: 4482
Joined: Wed Feb 19, 2014 10:30 am

Re: Evaluating SPIAs by implied return or life expectancy

Post by dcabler »

Chuck wrote: Wed May 31, 2023 8:25 am
petulant wrote: Wed May 31, 2023 7:49 am Now, another wrinkle is that many BH would consider getting a SPIA with a COLA. The simple Excel formulas don't work with a COLA; as far as I know, it would require building a larger spreadsheet and applying discount rates across cells.
Just run the same formulas using real dollars and a real interest rate, and interpret the numbers as such.
That would work if we still had SPIAS that were CPI adjusted. That's not what's available now - you now have to choose an annual % fixed COLA. That will of course reduce the payout and whether or not it keeps up with realized inflation going forward over the lifetime of the annuity is unknown.

Cheers
Chuck
Posts: 2505
Joined: Thu May 21, 2009 12:19 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by Chuck »

Makes sense. It's why there is a lot of support for delaying SS as long as possible, since it maximizes the only true inflation adjusted annuity most of us will have access to.
dcabler
Posts: 4482
Joined: Wed Feb 19, 2014 10:30 am

Re: Evaluating SPIAs by implied return or life expectancy

Post by dcabler »

Chuck wrote: Wed May 31, 2023 9:01 am Makes sense. It's why there is a lot of support for delaying SS as long as possible, since it maximizes the only true inflation adjusted annuity most of us will have access to.
Yep. And why true apples-to-apples comparison outside of nominal bonds vs. nominal SPIAS is kinda tough.

Cheers
User avatar
#Cruncher
Posts: 3975
Joined: Fri May 14, 2010 2:33 am
Location: New York City
Contact:

Re: Evaluating SPIAs by implied return or life expectancy

Post by #Cruncher »

petulant wrote: Wed May 31, 2023 7:49 amNow, another wrinkle is that many BH would consider getting a SPIA with a COLA. The simple Excel formulas don't work with a COLA; as far as I know, it would require building a larger spreadsheet and applying discount rates across cells.
We can handle this by adjusting the result of the Excel RATE function as follows along with illustrative values. [1]

Code: Select all

where
  nper =      16 = number of periods
   pmt =   7,884 = initial payment per period
    pv = 100,000 = present value
  grow =      2% = illustrative growth rate of pmt
return = (1 + RATE(nper, pmt,  -pv,     0, 0)) * (1 + grow) - 1
 4.93% = (1 + RATE(16,   7884, -100000, 0, 0)) * (1 + 2%)   - 1
To confirm this, here are the payments along with their discounted values. Scroll down to see that they sum to $100,000.

Code: Select all

Row      Col A       Col B       Col C  Formula in Column B
  2   Nbr pmts          16
  3        Pmt       7,884
  4         PV     100,000
  5  Pmts grow       2.00%
  6     Return       4.93%              =(1+RATE($B2,$B3,-$B4,0,0))*(1+B5)-1
  7     Period     Payment  Discounted

Code: Select all

  8          1       8,042       7,664
  9          2       8,203       7,450
 10          3       8,367       7,242
 11          4       8,534       7,039
 12          5       8,705       6,843
 13          6       8,879       6,652
 14          7       9,056       6,466
 15          8       9,237       6,285
 16          9       9,422       6,110
 17         10       9,611       5,939
 18         11       9,803       5,773
 19         12       9,999       5,612
 20         13      10,199       5,455
 21         14      10,403       5,303
 22         15      10,611       5,155
 23         16      10,823       5,011 [2]
 24      Total     149,891     100,000
  1. I chose 16 instead of the 15.82 in the original post to make a better illustration. But the formula for return also works for non-integer periods. For example:
    4.81% = (1 + RATE(15.82, 7884, -100000, 0, 0)) * 1.02 - 1
  2. Here is the calculation of the payment and its discounted value for period 16:

    Code: Select all

    10,823 =  7884 * 1.02   ^ 16
     5,011 = 10823 / 1.0493 ^ 16
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Evaluating SPIAs by implied return or life expectancy

Post by McQ »

petulant wrote: Wed May 31, 2023 7:49 am A Single Premium Immediate Annuity (SPIA) makes periodic payments for a guaranteed time period. While they can be structured for a number of time periods, the most popular form to discuss is the receipt of payments for the remainder of the purchaser's life. BH sometimes discuss SPIAs as one tool used to transform a portfolio of assets into safe retirement income.

Evaluating SPIAs can be difficult because a SPIA quote is stated as a payout rate that looks attractive if compared to the yields or expected returns on other financial products, but this payout rate is not equivalent to the yields on other financial products. A SPIA's payout rate includes the following components:
  • Investment returns
  • Return of principal
  • Mortality credits
Hence, what looks like a 9% payout rate is emphatically not a 9% return on investment. Consider a 15-year mortgage with a 4% interest rate. For each $100,000 in principal, that mortgage would have payments of almost $9,000 per year, or 9% of the starting balance. The actual cost of that debt is 4%, but the payments are 9%. That is what a SPIA with lifetime payments does: it is like a mortgage being paid by the insurance company to the retiree with a lower yield figure plus return of principal.

That means that every SPIA has an implied investment return figure, a period over which principal is returned, and a mortality/actuarial component.

Now, if we knew the time period of payments, the payment amount, and the starting balance, we could figure out the investment return implicit in the SPIA quote. But you don't know how long you live, and the insurance company is using an actuarial estimate. Or, given the payment amount, starting balance, and an interest rate, we could estimate the insurance company's estimate of your life expectancy. But we can't derive a fair life expectancy figure (the payment period) AND interest rate just from a SPIA payout rate.

Hence, the way to evaluate a SPIA payout rate is to get some independent estimate of one figure (the interest rate or life expectancy) and then solve for the other figure, then do vice versa. I'll work out an example.

I begin with a SPIA quote from immediateannuities.com for a 70-year-old female in Iowa paying $100,000. With no cash refund or period certain, the payout is $657 per month, or an annual payout rate of 7.884%.

To estimate the implied return, I could use an estimated life expectancy like the Social Security Administration's actuarial life table. Per that table, a 70-year-old female has a life expectancy of 15.82 years. So I can plug in the following Excel formula to get a close estimate of the implied investment return in the SPIA: =RATE(15.82,7884,-100000,0). In this case, the rate is 2.76%. Now, there are many people who believe the SSA actuarial table is not useful for this purpose because it covers the entire population, while those who purchase SPIAs tend to be wealthier, healthier, and longer-lived. So, I consulted the Bankrate Life Expectancy Calculator, answering various questions for my hypothetical 70-year-old woman in a reasonably healthy but not active way, resulting in a life expectancy of 18.5 more years. Using that for the formula, =RATE(18.5,7884,-100000,0), results in an implied return of 4.2%. So you can see health and life expectancy matter for evaluating the rate.

To gauge the insurance company's life expectancy estimate, we can use a proxy interest rate like the 20-year Treasury bond, which is currently yielding 4.06%. We plug that into the following Excel formula: =NPER(0.0406,7884,-100000,0). The result is 18.18 years. So, this is saying that if the insurance company starts with a rate of 4.06%, its quote is assuming a life expectancy of 18.18 years. As a purchaser, the hypothetical 70-year-old female in Iowa could evaluate whether she believes this figure is reasonable.

Now, another wrinkle is that many BH would consider getting a SPIA with a COLA. The simple Excel formulas don't work with a COLA; as far as I know, it would require building a larger spreadsheet and applying discount rates across cells. But hopefully this is a first step in helping people convert the payout rates on SPIAs into estimated returns.
Kudos, Petulant, on your post. I like how you step through the issues with simple formulae.

JoMoney later commented:
“I'll add in years past while interest rates were low, the payoff of a SPIA vs. withdrawing from a bond ladder over amortized over life expectancy was giving a much larger advantage to the SPIA... in more recent times that I've looked, having a bond ladder at current rates present a more competitive option”

I wrote a paper investigating exactly that question: when would a retiree have done better with bonds and when would a life annuity have provided higher present value. https://papers.ssrn.com/sol3/papers.cfm ... id=4160167

The paper will be tl;dr for 99% of forum members, but anyone who could follow the arithmetic in your post will have no trouble with the paper (most of the results are presented as charts). There’s quite a bit of bond market history woven in as well, as I took the test back to 1879.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
User avatar
nisiprius
Advisory Board
Posts: 52105
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Evaluating SPIAs by implied return or life expectancy

Post by nisiprius »

I have seen "money's worth" calculations in which they compared the price of SPIAs as a percentage of the theoretical price of an SPIA assuming life expectancy tables and investments in Treasury securities and the conclusion is always that they aren't bad. The present value of the future payouts of an SPIA are worth something in the ballpark of 90 to 95% of the premium.

The rate of return on investment is worth knowing to understand the whole situation, but it is also more or less irrelevant, because there is no way for an individual investor to do what an insurance company does--pool the risk from a large group of policyholders. Roughly speaking, in determining the payout from an SPIA, the insurance company is able to budget the payout on the average life expectancy of the group, while an individual has to budget on the basis of the maximum lifespan they wish to plan for.

Conceptually, I see it as a three-way choice.

1) Make a personal investment in low-risk bonds, and accept a very low withdrawal rate, e.g. 2.5% first-year payout. Low risk, low payouts, and probably considerable money left over at the end for someone to inherit.

2) Buy an SPIA, which represents an investment in low-risk bonds, and overall has very low risk; opt for a 3%/year increasing payout; get about double the payout (currently quoted at about $420/month on a $100,000 premium = about 5%/year first-year payout) and no money left at the end.*

3) Make a personal investment in a moderate-risk balanced portfolio of stocks and bonds, get about a 4%/year first-year payout, and almost certainly money left at the end--with the risk showing up in the form of a highly uncertain amount. The "4%" number is endlessly debated, but in my opinion there is strong evidence that 5% is too high--because that's exactly what Vanguard tried and failed to do in the Managed Payout Growth & Distribution fund.

1) Low payouts, low risk, and something left over for someone to inherit.
2) High payouts, low risk, and nothing left over for anyone to inherit.
3) High payouts, some risk, and a wildly unpredictable amount for someone to inherit

The big decision is between alternatives 2 and 3, and is mostly a decision involving risk tolerance and the desire to leave a legacy. Either way you get (very roughly) same amount of money available to spend yourself. That all the SPIA gives you, but it gives it to you with very little risk. Systematic portfolios don't give you any more money to spend yourself--by most calculations, noticeably less. And you are taking more risk in two ways. First, in the form of a small risk to yourself that you'll need to cut back if stocks perform poorly. Second, a high probability of leaving something to your kids, but risk in the form of a huge degree of uncertainty about how much it will be.

*Real-world annuities offer various options, such as a guarantee that there will be at least ten years of payments, in exchange for a lower payout.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
User avatar
Oicuryy
Posts: 1959
Joined: Thu Feb 22, 2007 9:29 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by Oicuryy »

The Society of Actuaries has an online annuity factor calculator.

https://afc.soa.org/

Ron
Money is fungible | Abbreviations and Acronyms
Topic Author
petulant
Posts: 3601
Joined: Thu Sep 22, 2016 1:09 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by petulant »

Oicuryy wrote: Wed May 31, 2023 6:07 pm The Society of Actuaries has an online annuity factor calculator.

https://afc.soa.org/

Ron
That one looks a bit complicated.
Wrench
Posts: 1046
Joined: Sun Apr 28, 2019 10:21 am

Re: Evaluating SPIAs by implied return or life expectancy

Post by Wrench »

nisiprius wrote: Wed May 31, 2023 5:54 pm I have seen "money's worth" calculations in which they compared the price of SPIAs as a percentage of the theoretical price of an SPIA assuming life expectancy tables and investments in Treasury securities and the conclusion is always that they aren't bad. The present value of the future payouts of an SPIA are worth something in the ballpark of 90 to 95% of the premium.

The rate of return on investment is worth knowing to understand the whole situation, but it is also more or less irrelevant, because there is no way for an individual investor to do what an insurance company does--pool the risk from a large group of policyholders. Roughly speaking, in determining the payout from an SPIA, the insurance company is able to budget the payout on the average life expectancy of the group, while an individual has to budget on the basis of the maximum lifespan they wish to plan for.

Conceptually, I see it as a three-way choice.

1) Make a personal investment in low-risk bonds, and accept a very low withdrawal rate, e.g. 2.5% first-year payout. Low risk, low payouts, and probably considerable money left over at the end for someone to inherit.

2) Buy an SPIA, which represents an investment in low-risk bonds, and overall has very low risk; opt for a 3%/year increasing payout; get about double the payout (currently quoted at about $420/month on a $100,000 premium = about 5%/year first-year payout) and no money left at the end.*

3) Make a personal investment in a moderate-risk balanced portfolio of stocks and bonds, get about a 4%/year first-year payout, and almost certainly money left at the end--with the risk showing up in the form of a highly uncertain amount. The "4%" number is endlessly debated, but in my opinion there is strong evidence that 5% is too high--because that's exactly what Vanguard tried and failed to do in the Managed Payout Growth & Distribution fund.

1) Low payouts, low risk, and something left over for someone to inherit.
2) High payouts, low risk, and nothing left over for anyone to inherit.
3) High payouts, some risk, and a wildly unpredictable amount for someone to inherit

The big decision is between alternatives 2 and 3, and is mostly a decision involving risk tolerance and the desire to leave a legacy. Either way you get (very roughly) same amount of money available to spend yourself. That all the SPIA gives you, but it gives it to you with very little risk. Systematic portfolios don't give you any more money to spend yourself--by most calculations, noticeably less. And you are taking more risk in two ways. First, in the form of a small risk to yourself that you'll need to cut back if stocks perform poorly. Second, a high probability of leaving something to your kids, but risk in the form of a huge degree of uncertainty about how much it will be.

*Real-world annuities offer various options, such as a guarantee that there will be at least ten years of payments, in exchange for a lower payout.
Thank you Nisi. A very clear summary of choices for retiree income! This ought to be in a BH wiki on Retirement Income (or maybe it is already?). I observe that BHs deeply discuss and argue over #3, with much less discussion or consideration for either #1 or #2. I would only add that these choices are not mutually exclusive. One can easily combine them, for example using #1 or #2 for basic necessary living income, and using #3 for discretionary income, which can better withstand highly variable, unpredictable returns.

Wrench
User avatar
nisiprius
Advisory Board
Posts: 52105
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Evaluating SPIAs by implied return or life expectancy

Post by nisiprius »

Oicuryy wrote: Wed May 31, 2023 6:07 pm The Society of Actuaries has an online annuity factor calculator.

https://afc.soa.org/

Ron
Can you use that to do a "money's worth" calculation on a real-world annuity? I don't understand what the "annuity factor" and the "duration" mean.

For example, if we take this online annuity estimate at face value... for a 65-year-old man living in Pennsylvania with a $100,000 investment and payments to start in 1 month...

Image

For example, if we blindly accept the tool's defaults and then set age = 65, benefit commencement = 65, primary annuitant gender = male, we get:

Image

What does that actually mean, and what percentage of money's worth does the online estimate represent?
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
User avatar
Stinky
Posts: 14088
Joined: Mon Jun 12, 2017 11:38 am
Location: Sweet Home Alabama

Re: Evaluating SPIAs by implied return or life expectancy

Post by Stinky »

nisiprius wrote: Wed May 31, 2023 5:54 pm Conceptually, I see it as a three-way choice.

1) Make a personal investment in low-risk bonds, and accept a very low withdrawal rate, e.g. 2.5% first-year payout. Low risk, low payouts, and probably considerable money left over at the end for someone to inherit.

2) Buy an SPIA, which represents an investment in low-risk bonds, and overall has very low risk; opt for a 3%/year increasing payout; get about double the payout (currently quoted at about $420/month on a $100,000 premium = about 5%/year first-year payout) and no money left at the end.*

3) Make a personal investment in a moderate-risk balanced portfolio of stocks and bonds, get about a 4%/year first-year payout, and almost certainly money left at the end--with the risk showing up in the form of a highly uncertain amount. The "4%" number is endlessly debated, but in my opinion there is strong evidence that 5% is too high--because that's exactly what Vanguard tried and failed to do in the Managed Payout Growth & Distribution fund.

1) Low payouts, low risk, and something left over for someone to inherit.
2) High payouts, low risk, and nothing left over for anyone to inherit.
3) High payouts, some risk, and a wildly unpredictable amount for someone to inherit
Very nicely stated.
Retired life insurance company financial executive who sincerely believes that ”It’s a GREAT day to be alive!”
Wrench
Posts: 1046
Joined: Sun Apr 28, 2019 10:21 am

Re: Evaluating SPIAs by implied return or life expectancy

Post by Wrench »

nisiprius wrote: Thu Jun 01, 2023 6:47 pm
Oicuryy wrote: Wed May 31, 2023 6:07 pm The Society of Actuaries has an online annuity factor calculator.

https://afc.soa.org/

Ron
Can you use that to do a "money's worth" calculation on a real-world annuity? I don't understand what the "annuity factor" and the "duration" mean.

For example, if we take this online annuity estimate at face value... for a 65-year-old man living in Pennsylvania with a $100,000 investment and payments to start in 1 month...

Image

For example, if we blindly accept the tool's defaults and then set age = 65, benefit commencement = 65, primary annuitant gender = male, we get:

Image

What does that actually mean, and what percentage of money's worth does the online estimate represent?
I am NOT an actuary, but my understanding is the annuity factor determines the present value of a stream of future payouts at a given discount rate. So, for the above example, if you wanted income of 12*$607=$7284 per year, you would need 13.4351*$7284 = $97861.26 today. So your estimate of the discount rate is apparently a reasonable guestimate of what the insurance company may have used in the quote from immediateannuities web site. (Maybe the company uses the extra~$2.1K to pay the agent's commission??). The long term AAA-rated commercial bond rates today are on the order of 4.5% so the company (crudely) charges ~0.5% "fee" as their profit.
I was not aware of "Duration" used in this context, but the same web site says it is
"The approximate percentage change in annuity factor per 100 basis point change in discount rate. Duration of 8.2 means that for discount rate decrease of 100 basis points, the annuity factor will increase by approximately 8.2%. There are a variety of types of duration; AFC calculates effective duration."
So, essentially the same definition as duration of bonds.

Wrench
User avatar
nisiprius
Advisory Board
Posts: 52105
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Evaluating SPIAs by implied return or life expectancy

Post by nisiprius »

Wrench wrote: Thu Jun 01, 2023 7:47 pm I am NOT an actuary, but my understanding is the annuity factor determines the present value of a stream of future payouts at a given discount rate. So, for the above example, if you wanted income of 12*$607=$7284 per year, you would need 13.4351*$7284 = $97861.26 today...
Thank you very much. So, on the face of it, under the default assumptions in the SOA tool and the estimate from the annuity broker, an annuity that costs $100,000 is legitimately worth $97,861.26, so the annuity is providing over 97% of your money's worth--or to put it another way, the insurance company is only charging a "load" of a few percent.

Of course it's very sensitive to that discount rate. If the discount rate is 5%, the annuity factor drops to 12.2650, and the $100,000 annuity is only worth 12.2650 * 12 * $607 = $89,272.
Last edited by nisiprius on Thu Jun 01, 2023 8:08 pm, edited 1 time in total.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
Wrench
Posts: 1046
Joined: Sun Apr 28, 2019 10:21 am

Re: Evaluating SPIAs by implied return or life expectancy

Post by Wrench »

One more thing about the SOA calculator: it allows one to put in an annual (fixed) COLA. So one can see the cost of inflation protection! Using Nisi's example, increasing the COLA from 0 to 3%, increases the annuity factor from 13.4351 to 18.1702, an ~35% increase. Since 3% is the 100 year average of actual measured inflation, this is probably a reasonable estimate for what a CPI-U adjusted annuity would cost over a nominal annuity. (This is consistent with UK annuity rates where annuities with inflation adjustments are available, and with my own experience buying a CPI-U annuity in the US when they were available). Also note that this is for a 65 year old. The additional cost for the for inflation coverage will increase for younger annuitants, and decrease for older annuitants. For example, at age 80 the premium only goes up ~17% instead of 35%. This makes sense because there are fewer years for the inflation to act on an 80 year old compared to a 65 year old.

All of which helps explain why there aren't any CPI-U inflation adjusted annuities today. How many consumers would be willing to pay an extra 35% for something (inflation) that may or may not happen? We know the answer - not enough to make it worth the insurance company's effort to offer the product. But, if we have many years of high single digit or double digit inflation maybe that will change...

Wrench
User avatar
Stinky
Posts: 14088
Joined: Mon Jun 12, 2017 11:38 am
Location: Sweet Home Alabama

Re: Evaluating SPIAs by implied return or life expectancy

Post by Stinky »

Wrench wrote: Thu Jun 01, 2023 7:47 pm
nisiprius wrote: Thu Jun 01, 2023 6:47 pm
Oicuryy wrote: Wed May 31, 2023 6:07 pm The Society of Actuaries has an online annuity factor calculator.

https://afc.soa.org/

Ron
Can you use that to do a "money's worth" calculation on a real-world annuity? I don't understand what the "annuity factor" and the "duration" mean.

For example, if we take this online annuity estimate at face value... for a 65-year-old man living in Pennsylvania with a $100,000 investment and payments to start in 1 month...

Image

For example, if we blindly accept the tool's defaults and then set age = 65, benefit commencement = 65, primary annuitant gender = male, we get:

Image

What does that actually mean, and what percentage of money's worth does the online estimate represent?
I am NOT an actuary, but my understanding is the annuity factor determines the present value of a stream of future payouts at a given discount rate. So, for the above example, if you wanted income of 12*$607=$7284 per year, you would need 13.4351*$7284 = $97861.26 today. So your estimate of the discount rate is apparently a reasonable guestimate of what the insurance company may have used in the quote from immediateannuities web site. (Maybe the company uses the extra~$2.1K to pay the agent's commission??). The long term AAA-rated commercial bond rates today are on the order of 4.5% so the company (crudely) charges ~0.5% "fee" as their profit.
I was not aware of "Duration" used in this context, but the same web site says it is
"The approximate percentage change in annuity factor per 100 basis point change in discount rate. Duration of 8.2 means that for discount rate decrease of 100 basis points, the annuity factor will increase by approximately 8.2%. There are a variety of types of duration; AFC calculates effective duration."
So, essentially the same definition as duration of bonds.

Wrench
I am an actuary.

I’ve not used this particular tool on the SOA website, but I’m familiar with the concepts and the terms.

Poster “Wrench” actually did a very good job in his description of the annuity factor. It is a factor that can be used to determine the present value of a stream of annual life-contingent payments. His calculation of the value of the SPIA payment, given the input parameters, was the way it would be done. That is, (monthly SPIA payment) time 12 times annuity factor.

The problem with we mere mortals using a tool like this is that we have the input parameters all wrong. I’m not familiar with the “default” mortality table (“PRI 2012….”), but it appears to be a mortality table for a corporate pension plan rather than a individual immediate annuity population. The 4% interest rate used as a “default” is simply a plug number. There’s no provision in this exercise for insurer actual earned rates, default charges, risk charges, expense, taxes, capital charges, and the zillion other things that go into pricing a SPIA.

In other words, the “annuity factor” happens to pretty closely replicate the price for the SPIA quoted from
Immediateannuities.com. But it’s not because a tool like this is used for actual annuity pricing.
Retired life insurance company financial executive who sincerely believes that ”It’s a GREAT day to be alive!”
User avatar
nisiprius
Advisory Board
Posts: 52105
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Evaluating SPIAs by implied return or life expectancy

Post by nisiprius »

Wrench wrote: Thu Jun 01, 2023 8:06 pm...All of which helps explain why there aren't any CPI-U inflation adjusted annuities today. How many consumers would be willing to pay an extra 35% for something (inflation) that may or may not happen?
Well, I was. And based on the discussions a couple of years ago when American General applied too small an inflation adjustment, I wasn't alone.
We know the answer - not enough to make it worth the insurance company's effort to offer the product. But, if we have many years of high single digit or double digit inflation maybe that will change...

Wrench
The power of money illusion is strange. The reason why you want an inflation adjustment is that you intuit, correctly, that inflation will be a huge problem and will greatly cut the purchasing power of the payouts. So you would think that since an inflation-indexed, or 3%-annual-increasing payout is expected to pay out a lot more, you would expect it to cost a lot more. But apparently people were experiencing sticker shock.
Last edited by nisiprius on Thu Jun 01, 2023 8:19 pm, edited 1 time in total.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
User avatar
nisiprius
Advisory Board
Posts: 52105
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Evaluating SPIAs by implied return or life expectancy

Post by nisiprius »

Stinky wrote: Thu Jun 01, 2023 8:10 pm ...The problem with we mere mortals using a tool like this is that we have the input parameters all wrong. I’m not familiar with the “default” mortality table (“PRI 2012….”), but it appears to be a mortality table for a corporate pension plan rather than a individual immediate annuity population. The 4% interest rate used as a “default” is simply a plug number. There’s no provision in this exercise for insurer actual earned rates, default charges, risk charges, expense, taxes, capital charges, and the zillion other things that go into pricing a SPIA.

In other words, the “annuity factor” happens to pretty closely replicate the price for the SPIA quoted from
Immediateannuities.com. But it’s not because a tool like this is used for actual annuity pricing...
Of course not, and they make you check a checkbox that says you understand this. But I think it's valuable as a reality check on annuity prices.

Before disclosure of medical loss ratios was often mandated, it was not rare to find health insurance plans with 50% medical loss ratios, i.e. the insurer was charging 2X what the insurance was really "worth." That doesn't seem to be the case with SPIAs, where the insurer's profit margin according to studies I've read is in the 5-10% ballpark. But if you can shed more light on this I'm all ears.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
Wrench
Posts: 1046
Joined: Sun Apr 28, 2019 10:21 am

Re: Evaluating SPIAs by implied return or life expectancy

Post by Wrench »

nisiprius wrote: Thu Jun 01, 2023 8:02 pm
Wrench wrote: Thu Jun 01, 2023 7:47 pm I am NOT an actuary, but my understanding is the annuity factor determines the present value of a stream of future payouts at a given discount rate. So, for the above example, if you wanted income of 12*$607=$7284 per year, you would need 13.4351*$7284 = $97861.26 today...
Thank you very much. So, on the face of it, under the default assumptions in the SOA tool and the estimate from the annuity broker, an annuity that costs $100,000 is legitimately worth $97,861.26, so the annuity is providing over 97% of your money's worth--or to put it another way, the insurance company is only charging a "load" of a few percent.

Of course it's very sensitive to that discount rate. If the discount rate is 5%, the annuity factor drops to 12.2650, and the $100,000 annuity is only worth 12.2650 * 12 * $607 = $89,272.
But who knows how they REALLY do it? If you use the Mortality table for Top 25% healthy annuitant, the factor for your 65 yo male with the all the other parameters the same is 14.4576. That means you are getting a real bargain because it only costs you $100K for 12*607*14.4576 = $105309 it is costing the company! Maybe they are OK with that because they make it up on the interest rate spread? Or they use a different discount rate? Only the actuaries who do this for a living really know for sure, and they probably couldn't tell us because it is proprietary.

Wrench
Wrench
Posts: 1046
Joined: Sun Apr 28, 2019 10:21 am

Re: Evaluating SPIAs by implied return or life expectancy

Post by Wrench »

nisiprius wrote: Thu Jun 01, 2023 8:13 pm
Wrench wrote: Thu Jun 01, 2023 8:06 pm...All of which helps explain why there aren't any CPI-U inflation adjusted annuities today. How many consumers would be willing to pay an extra 35% for something (inflation) that may or may not happen?
Well, I was. And based on the discussions a couple of years ago when American General applied too small an inflation adjustment, I wasn't alone.
We know the answer - not enough to make it worth the insurance company's effort to offer the product. But, if we have many years of high single digit or double digit inflation maybe that will change...

Wrench
The power of money illusion is strange. The reason why you want an inflation adjustment is that you intuit, correctly, that inflation will be a huge problem and will greatly cut the purchasing power of the payouts. So you would think that since an inflation-indexed, or 3%-annual-increasing payout is expected to pay out a lot more, you would expect it to cost a lot more. But apparently people were experiencing sticker shock.
I think it is way more complicated than that because whether it makes sense or not depends on how long you will live. If you die in the first few years, even with high inflation it doesn't pay off to get inflation protection. OTOH, if you live 30 more years, even with 2% inflation it's a huge advantage. A recent WSJ article on life expectancy suggested that most people underestimate how long they will live by a number of years. So, at age 65 if you think you will only live to 75 or 80, are you going to pay 35% extra for inflation protection? You and I probably would (and we did!) but again, apparently not that many folks would. A real life example - we helped my 89 yo recently widowed MIL purchase an SPIA. We intentionally did NOT look at inflation protection because over her expected lifetime we did not think it was worth it. The probability she will live 15 years is close to nil, and although she might make 10 more years (she is VERY healthy!) the potential impact of inflation over that time frame was not sufficient for us to ante up the additional premium.

Wrench
User avatar
nisiprius
Advisory Board
Posts: 52105
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Evaluating SPIAs by implied return or life expectancy

Post by nisiprius »

Going back to the original poster's question... in principle, can you get the effective rate of return on an SPIA by calculating the annuity factor, and adjusting the discount rate in the SOA tool until it gives the same annuity factor? Then the discount rate is the effective internal rate of return for the SPIA.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.
User avatar
Stinky
Posts: 14088
Joined: Mon Jun 12, 2017 11:38 am
Location: Sweet Home Alabama

Re: Evaluating SPIAs by implied return or life expectancy

Post by Stinky »

nisiprius wrote: Thu Jun 01, 2023 8:18 pm
Stinky wrote: Thu Jun 01, 2023 8:10 pm ...The problem with we mere mortals using a tool like this is that we have the input parameters all wrong. I’m not familiar with the “default” mortality table (“PRI 2012….”), but it appears to be a mortality table for a corporate pension plan rather than a individual immediate annuity population. The 4% interest rate used as a “default” is simply a plug number. There’s no provision in this exercise for insurer actual earned rates, default charges, risk charges, expense, taxes, capital charges, and the zillion other things that go into pricing a SPIA.

In other words, the “annuity factor” happens to pretty closely replicate the price for the SPIA quoted from
Immediateannuities.com. But it’s not because a tool like this is used for actual annuity pricing...
Of course not, and they make you check a checkbox that says you understand this. But I think it's valuable as a reality check on annuity prices.

Before disclosure of medical loss ratios was often mandated, it was not rare to find health insurance plans with 50% medical loss ratios, i.e. the insurer was charging 2X what the insurance was really "worth." That doesn't seem to be the case with SPIAs, where the insurer's profit margin according to studies I've read is in the 5-10% ballpark. But if you can shed more light on this I'm all ears.
Yes, a tool like this can be useful, not just as a check on annuity prices, but also to see the relative difference in payouts based on changes in assumptions. For example, how much is the factor affected by a 1% change in discount rates? How about an inflation factor, female instead of male, etc.?

I know that most life insurance companies price for a high single digit or low double digit after tax return on invested capital, rsther than basing pricing on a percent of premium. Your conjecture of 5% of premium is in the range.
Retired life insurance company financial executive who sincerely believes that ”It’s a GREAT day to be alive!”
User avatar
Stinky
Posts: 14088
Joined: Mon Jun 12, 2017 11:38 am
Location: Sweet Home Alabama

Re: Evaluating SPIAs by implied return or life expectancy

Post by Stinky »

nisiprius wrote: Thu Jun 01, 2023 9:03 pm Going back to the original poster's question... in principle, can you get the effective rate of return on an SPIA by calculating the annuity factor, and adjusting the discount rate in the SOA tool until it gives the same annuity factor? Then the discount rate is the effective internal rate of return for the SPIA.
Yes, I think you’re close. But——

I’m not familiar with the mortality tables used in the tool. They appear to be based on defined benefit pension plan experience, rather than individual life SPIA experience.

I’d expect that pension plan mortality rates would be higher than SPIA mortality rates, because of the self-selection of SPIA buyers.

Therefore, the “annuity factor” (and therefore, the single premium price) for a given amount of monthly income should be higher for a SPIA than a pension payout.

Maybe this can be adjusted for by using the lowest mortality rates in the tool. I see some tables labeled “2014 top 25%”. That might refer to the top 25% of salaries, which might have lower mortality rates due to socioeconomic factors. It that’s just a guess on my part.
Retired life insurance company financial executive who sincerely believes that ”It’s a GREAT day to be alive!”
Topic Author
petulant
Posts: 3601
Joined: Thu Sep 22, 2016 1:09 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by petulant »

#Cruncher wrote: Wed May 31, 2023 3:59 pm
petulant wrote: Wed May 31, 2023 7:49 amNow, another wrinkle is that many BH would consider getting a SPIA with a COLA. The simple Excel formulas don't work with a COLA; as far as I know, it would require building a larger spreadsheet and applying discount rates across cells.
We can handle this by adjusting the result of the Excel RATE function as follows along with illustrative values. ...
Thanks for posting. I just had a minute today to review. I think it's a good first approximation, and I'll add it to an edit in the OP. However, it is important to note that this method is only precise under specific payment timing assumptions: payments are annual, they begin one year after entering the contract, and the first COLA applies to the first payment one year from purchase. I believe that under different assumptions, the result could be as much as 30 bps off. That's not huge, and it's a good first approximation, but I wanted to make that caveat.
Topic Author
petulant
Posts: 3601
Joined: Thu Sep 22, 2016 1:09 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by petulant »

Stinky wrote: Thu Jun 01, 2023 8:10 pm
Wrench wrote: Thu Jun 01, 2023 7:47 pm
nisiprius wrote: Thu Jun 01, 2023 6:47 pm
Oicuryy wrote: Wed May 31, 2023 6:07 pm The Society of Actuaries has an online annuity factor calculator.

https://afc.soa.org/

Ron
Can you use that to do a "money's worth" calculation on a real-world annuity? I don't understand what the "annuity factor" and the "duration" mean.

For example, if we take this online annuity estimate at face value... for a 65-year-old man living in Pennsylvania with a $100,000 investment and payments to start in 1 month...

Image

For example, if we blindly accept the tool's defaults and then set age = 65, benefit commencement = 65, primary annuitant gender = male, we get:

Image

What does that actually mean, and what percentage of money's worth does the online estimate represent?
I am NOT an actuary, but my understanding is the annuity factor determines the present value of a stream of future payouts at a given discount rate. So, for the above example, if you wanted income of 12*$607=$7284 per year, you would need 13.4351*$7284 = $97861.26 today. So your estimate of the discount rate is apparently a reasonable guestimate of what the insurance company may have used in the quote from immediateannuities web site. (Maybe the company uses the extra~$2.1K to pay the agent's commission??). The long term AAA-rated commercial bond rates today are on the order of 4.5% so the company (crudely) charges ~0.5% "fee" as their profit.
I was not aware of "Duration" used in this context, but the same web site says it is
"The approximate percentage change in annuity factor per 100 basis point change in discount rate. Duration of 8.2 means that for discount rate decrease of 100 basis points, the annuity factor will increase by approximately 8.2%. There are a variety of types of duration; AFC calculates effective duration."
So, essentially the same definition as duration of bonds.

Wrench
I am an actuary.

I’ve not used this particular tool on the SOA website, but I’m familiar with the concepts and the terms.

Poster “Wrench” actually did a very good job in his description of the annuity factor. It is a factor that can be used to determine the present value of a stream of annual life-contingent payments. His calculation of the value of the SPIA payment, given the input parameters, was the way it would be done. That is, (monthly SPIA payment) time 12 times annuity factor.

The problem with we mere mortals using a tool like this is that we have the input parameters all wrong. I’m not familiar with the “default” mortality table (“PRI 2012….”), but it appears to be a mortality table for a corporate pension plan rather than a individual immediate annuity population. The 4% interest rate used as a “default” is simply a plug number. There’s no provision in this exercise for insurer actual earned rates, default charges, risk charges, expense, taxes, capital charges, and the zillion other things that go into pricing a SPIA.

In other words, the “annuity factor” happens to pretty closely replicate the price for the SPIA quoted from
Immediateannuities.com. But it’s not because a tool like this is used for actual annuity pricing.
I think the issue with the tool is that it still has the problem identified in OP...if you get a SPIA quote, you can assume a particular life expectancy (hence the mortality tables) and then get an implied rate of return, or you can assume a rate of return and use trial and error to figure out the life expectancy (switching mortality tables until you get the right one). While I imagine a power user could get some good results using the calculator, I think the formulas I posted (along with #Cruncher's proposal for COLA SPIAs) are better for generic users just trying to convert the SPIA payout rate into a rate of return. That's my biggest concern, since there still seems to be a lot of users who are confusing the payout rate with an interest rate, which is dreadfully wrong.
User avatar
Taylor Larimore
Posts: 32839
Joined: Tue Feb 27, 2007 7:09 pm
Location: Miami FL

Re: Evaluating SPIAs by implied return or life expectancy

Post by Taylor Larimore »

Bogleheads:

One of the nice things about buying SPIAs (Single Premium Immediate Annuities) is that they are nearly identical (you pay a one-time premium and they guarantee a life income--like a pension or social security). Because they are nearly identical their premiums are
very competitive. In my view, there is not much more to evaluate. I own two and they are the best investments we ever made.

Best wishes.
Taylor
Jack Bogle's Words of Wisdom:
"I probably prefer an immediate annuity which begins paying you right away."
"Simplicity is the master key to financial success." -- Jack Bogle
User avatar
Lawrence of Suburbia
Posts: 691
Joined: Mon Aug 08, 2022 12:04 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by Lawrence of Suburbia »

Taylor Larimore wrote: Sat Jun 03, 2023 11:56 am Bogleheads:

One of the nice things about buying SPIAs (Single Premium Immediate Annuities) is that they are nearly identical (you pay a one-time premium and they guarantee a life income--like a pension or social security). Because they are nearly identical their premiums are very competitive. In my view, there is not much more to evaluate. I own two and they are the best investments we ever made.

Best wishes.
Taylor
[...]
This, I think.

Apart from caculating how the SPIA fits in with one's overall financial plan, of course. It's plainly not right for some people -- those who plainly don't have enough saved to retire, simply can't afford one; and those whose WR is less than 2% (say) just don't need one, probably.

When my monthly SPIA check hits my bank account and goes towards my bills, I don't worry about how much is my principal being returned, how much is from bonds held by the insurance company, or (well off in the future
in my case) how much is mortality credits.
74% VTHRX/8% DODWX/12% TIAA Traditional/6% SWVXX
StillGoing
Posts: 359
Joined: Mon Nov 04, 2019 3:43 am
Location: U.K.

Re: Evaluating SPIAs by implied return or life expectancy

Post by StillGoing »

As someone who has poked under the bonnet of annuity payout rates, having attempted to match historical payout rates in both the US and UK in order to generate a time series of payout rates using historical interest rates and modern mortality statistics for backtesting purposes, this is an interesting thread. As already identified, there are 3 factors in determining the payout rate:

1) The return on investments
2) Mortality statistics
3) Fees

That different insurance companies end up using different values for each of these is exemplified by the variation in quoted prices, e.g. in the Summer 2021 edition of annuity shopper (https://www.immediateannuities.com/pdfs ... .pdf?arx=d) (this no longer seems to be updated - does anyone know of an equivalent where many quotes are given rather than just the average or maximum?), the average payout rate for a level annuity for a 65 year old male (life only) was $483 per month (per $100k premium). However, the minimum quote was $458 and the maximum quote was $503, a range of roughly plus/minus 5% of the average value. Historically, the range has been as large as +/-10% (e.g., see Figure 1 in Warshawsky, Private Annuity Markets in the United States: 1919-1984 - there are some nice calculations of load factors in that paper too).

The range of investments (UK based and for the rather specialised situation of pension buy outs) is alluded to in https://www.wtwco.com/en-gb/insights/20 ... der-invest - particularly interesting is the illiquidity premium gained on some of the investments and the emphasis on cash flow.

Mortality statistics are fairly well known (but there are many versions - the ones used in the UK by insurance companies are not publicly available), but presumably small providers (or for niche SPIA products) are priced more conservatively because they have a smaller population sample.

So, those offering the better deals have either taken more risk with their investments (or been able to secure access to "better" investments), with their mortality statistics (or have access to a broader population range), or are minimising their fees (either through productivity enhancements or by taking a reduction in profits).

cheers
StillGoing
Derpalator
Posts: 215
Joined: Sun Nov 15, 2015 1:52 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by Derpalator »

StillGoing wrote: Sun Jun 04, 2023 1:55 am As someone who has poked under the bonnet of annuity payout rates, having attempted to match historical payout rates in both the US and UK in order to generate a time series of payout rates using historical interest rates and modern mortality statistics for backtesting purposes, this is an interesting thread. As already identified, there are 3 factors in determining the payout rate:

1) The return on investments
2) Mortality statistics
3) Fees

That different insurance companies end up using different values for each of these is exemplified by the variation in quoted prices, e.g. in the Summer 2021 edition of annuity shopper (https://www.immediateannuities.com/pdfs ... .pdf?arx=d) (this no longer seems to be updated - does anyone know of an equivalent where many quotes are given rather than just the average or maximum?), the average payout rate for a level annuity for a 65 year old male (life only) was $483 per month (per $100k premium). However, the minimum quote was $458 and the maximum quote was $503, a range of roughly plus/minus 5% of the average value. Historically, the range has been as large as +/-10% (e.g., see Figure 1 in Warshawsky, Private Annuity Markets in the United States: 1919-1984 - there are some nice calculations of load factors in that paper too).

The range of investments (UK based and for the rather specialised situation of pension buy outs) is alluded to in https://www.wtwco.com/en-gb/insights/20 ... der-invest - particularly interesting is the illiquidity premium gained on some of the investments and the emphasis on cash flow.

Mortality statistics are fairly well known (but there are many versions - the ones used in the UK by insurance companies are not publicly available), but presumably small providers (or for niche SPIA products) are priced more conservatively because they have a smaller population sample.

So, those offering the better deals have either taken more risk with their investments (or been able to secure access to "better" investments), with their mortality statistics (or have access to a broader population range), or are minimising their fees (either through productivity enhancements or by taking a reduction in profits).

cheers
StillGoing
Many on this forum have recommended blueprint https://www.blueprintincome.com/
StillGoing
Posts: 359
Joined: Mon Nov 04, 2019 3:43 am
Location: U.K.

Re: Evaluating SPIAs by implied return or life expectancy

Post by StillGoing »

Lawrence of Suburbia wrote: Sat Jun 03, 2023 10:46 pm
Taylor Larimore wrote: Sat Jun 03, 2023 11:56 am Bogleheads:

One of the nice things about buying SPIAs (Single Premium Immediate Annuities) is that they are nearly identical (you pay a one-time premium and they guarantee a life income--like a pension or social security). Because they are nearly identical their premiums are very competitive. In my view, there is not much more to evaluate. I own two and they are the best investments we ever made.

Best wishes.
Taylor
[...]
This, I think.

Apart from caculating how the SPIA fits in with one's overall financial plan, of course. It's plainly not right for some people -- those who plainly don't have enough saved to retire, simply can't afford one; and those whose WR is less than 2% (say) just don't need one, probably.

When my monthly SPIA check hits my bank account and goes towards my bills, I don't worry about how much is my principal being returned, how much is from bonds held by the insurance company, or (well off in the future
in my case) how much is mortality credits.
I agree, from a retiree point of view how an annuity might fit into a retirement plan is the question rather than what is under the hood or value for money and the comparison should probably be with the alternatives which are (leaving aside social security/state pension and DB pensions)

1) Withdrawals from a stock/bond portfolio
2) Payments from a non-rolling bond ladder

The comparison should be appropriate too. For example, in the UK (I happen to have some recent figures available - the quotes were generated today) SPIAs are currently available such that (all for 65 yo with no guarantee period and quoted to nearest 10 basis points)

Code: Select all

Single life		Level	7.1%
			RPI	4.5%
Joint (50%) 		Level	6.6%
			RPI	4.0%
Joint (100%)		Level	6.1%
                   	RPI 	3.6%
While the SAFEMAX (in %) for a stock and bond portfolio over a 35 year period (i.e. to 100yo for a 65yo retiree). All assume 50% bonds and 50% cash in the fixed income component and use asset returns and cpi/rpi from macrohistory.net. For the level case, the nominal payout rate is fixed.

Code: Select all

		Stock allocation (%)
		40	60	80
Level		4.6	4.8	5.0
RPI		2.3	2.7	2.9
And constructing a 35 year ladder from UK Government bonds gives ladder payout rates (in %) of (I've used 15 year spot rates for 1 June 2023 downloaded from https://www.bankofengland.co.uk/statistics/yield-curves)

Code: Select all

Level	5.4
RPI	3.3
Costs for ladder and portfolio are ignored as are taxes.

In each case (level or RPI), the annuity (even joint with 100% survivor benefits) currently gives a better income rate than either ladder or portfolio (this would not have necessarily been the case 2 year ago - e.g. in July 2021, the payouts on single life annuities were 5.0% and 2.8% for level and RPI, respectively - much less attractive compared to the portfolio SAFEMAX).

However, legacy motives (in the event of early death) are a consideration as well as how well other sources of guaranteed income cover spending.

My apologies for a longish post probably only obliquely related to the topic.

cheers
StillGoing
Last edited by StillGoing on Sun Jun 04, 2023 4:12 am, edited 1 time in total.
StillGoing
Posts: 359
Joined: Mon Nov 04, 2019 3:43 am
Location: U.K.

Re: Evaluating SPIAs by implied return or life expectancy

Post by StillGoing »

Derpalator wrote: Sun Jun 04, 2023 2:34 am
StillGoing wrote: Sun Jun 04, 2023 1:55 am As someone who has poked under the bonnet of annuity payout rates, having attempted to match historical payout rates in both the US and UK in order to generate a time series of payout rates using historical interest rates and modern mortality statistics for backtesting purposes, this is an interesting thread. As already identified, there are 3 factors in determining the payout rate:

1) The return on investments
2) Mortality statistics
3) Fees

That different insurance companies end up using different values for each of these is exemplified by the variation in quoted prices, e.g. in the Summer 2021 edition of annuity shopper (https://www.immediateannuities.com/pdfs ... .pdf?arx=d) (this no longer seems to be updated - does anyone know of an equivalent where many quotes are given rather than just the average or maximum?), the average payout rate for a level annuity for a 65 year old male (life only) was $483 per month (per $100k premium). However, the minimum quote was $458 and the maximum quote was $503, a range of roughly plus/minus 5% of the average value. Historically, the range has been as large as +/-10% (e.g., see Figure 1 in Warshawsky, Private Annuity Markets in the United States: 1919-1984 - there are some nice calculations of load factors in that paper too).

The range of investments (UK based and for the rather specialised situation of pension buy outs) is alluded to in https://www.wtwco.com/en-gb/insights/20 ... der-invest - particularly interesting is the illiquidity premium gained on some of the investments and the emphasis on cash flow.

Mortality statistics are fairly well known (but there are many versions - the ones used in the UK by insurance companies are not publicly available), but presumably small providers (or for niche SPIA products) are priced more conservatively because they have a smaller population sample.

So, those offering the better deals have either taken more risk with their investments (or been able to secure access to "better" investments), with their mortality statistics (or have access to a broader population range), or are minimising their fees (either through productivity enhancements or by taking a reduction in profits).

cheers
StillGoing
Many on this forum have recommended blueprint https://www.blueprintincome.com/
Many thanks - however, they only appear to give 'best' rates rather than a list (like annuity shopper did) - I'm not in the US, so cannot ask for an actual quote (I'm only interested in the US from a research point of view).

cheers
StillGoing
Wrench
Posts: 1046
Joined: Sun Apr 28, 2019 10:21 am

Re: Evaluating SPIAs by implied return or life expectancy

Post by Wrench »

StillGoing wrote: Sun Jun 04, 2023 4:03 am
Lawrence of Suburbia wrote: Sat Jun 03, 2023 10:46 pm
Taylor Larimore wrote: Sat Jun 03, 2023 11:56 am Bogleheads:

One of the nice things about buying SPIAs (Single Premium Immediate Annuities) is that they are nearly identical (you pay a one-time premium and they guarantee a life income--like a pension or social security). Because they are nearly identical their premiums are very competitive. In my view, there is not much more to evaluate. I own two and they are the best investments we ever made.

Best wishes.
Taylor
[...]
This, I think.

Apart from caculating how the SPIA fits in with one's overall financial plan, of course. It's plainly not right for some people -- those who plainly don't have enough saved to retire, simply can't afford one; and those whose WR is less than 2% (say) just don't need one, probably.

When my monthly SPIA check hits my bank account and goes towards my bills, I don't worry about how much is my principal being returned, how much is from bonds held by the insurance company, or (well off in the future
in my case) how much is mortality credits.
I agree, from a retiree point of view how an annuity might fit into a retirement plan is the question rather than what is under the hood or value for money and the comparison should probably be with the alternatives which are (leaving aside social security/state pension and DB pensions)

1) Withdrawals from a stock/bond portfolio
2) Payments from a non-rolling bond ladder

The comparison should be appropriate too. For example, in the UK (I happen to have some recent figures available - the quotes were generated today) SPIAs are currently available such that (all for 65 yo with no guarantee period and quoted to nearest 10 basis points)

Code: Select all

Single life		Level	7.1%
			RPI	4.5%
Joint (50%) 		Level	6.6%
			RPI	4.0%
Joint (100%)		Level	6.1%
                   	RPI 	3.6%
While the SAFEMAX (in %) for a stock and bond portfolio over a 35 year period (i.e. to 100yo for a 65yo retiree). All assume 50% bonds and 50% cash in the fixed income component and use asset returns and cpi/rpi from macrohistory.net. For the level case, the nominal payout rate is fixed.

Code: Select all

		Stock allocation (%)
		40	60	80
Level		4.6	4.8	5.0
RPI		2.3	2.7	2.9
And constructing a 35 year ladder from UK Government bonds gives ladder payout rates (in %) of (I've used 15 year spot rates for 1 June 2023 downloaded from https://www.bankofengland.co.uk/statistics/yield-curves)

Code: Select all

Level	5.4
RPI	3.3
Costs for ladder and portfolio are ignored as are taxes.

In each case (level or RPI), the annuity (even joint with 100% survivor benefits) currently gives a better income rate than either ladder or portfolio (this would not have necessarily been the case 2 year ago - e.g. in July 2021, the payouts on single life annuities were 5.0% and 2.8% for level and RPI, respectively - much less attractive compared to the portfolio SAFEMAX).

However, legacy motives (in the event of early death) are a consideration as well as how well other sources of guaranteed income cover spending.

My apologies for a longish post probably only obliquely related to the topic.

cheers
StillGoing
Wow! I love your posts. They are always on point, backed up by real data. Thank you for your contributions to BH!

Wrench
Topic Author
petulant
Posts: 3601
Joined: Thu Sep 22, 2016 1:09 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by petulant »

StillGoing wrote: Sun Jun 04, 2023 1:55 am As someone who has poked under the bonnet of annuity payout rates, having attempted to match historical payout rates in both the US and UK in order to generate a time series of payout rates using historical interest rates and modern mortality statistics for backtesting purposes, this is an interesting thread. As already identified, there are 3 factors in determining the payout rate:

1) The return on investments
2) Mortality statistics
3) Fees

That different insurance companies end up using different values for each of these is exemplified by the variation in quoted prices, e.g. in the Summer 2021 edition of annuity shopper (https://www.immediateannuities.com/pdfs ... .pdf?arx=d) (this no longer seems to be updated - does anyone know of an equivalent where many quotes are given rather than just the average or maximum?), the average payout rate for a level annuity for a 65 year old male (life only) was $483 per month (per $100k premium). However, the minimum quote was $458 and the maximum quote was $503, a range of roughly plus/minus 5% of the average value. Historically, the range has been as large as +/-10% (e.g., see Figure 1 in Warshawsky, Private Annuity Markets in the United States: 1919-1984 - there are some nice calculations of load factors in that paper too).

The range of investments (UK based and for the rather specialised situation of pension buy outs) is alluded to in https://www.wtwco.com/en-gb/insights/20 ... der-invest - particularly interesting is the illiquidity premium gained on some of the investments and the emphasis on cash flow.

Mortality statistics are fairly well known (but there are many versions - the ones used in the UK by insurance companies are not publicly available), but presumably small providers (or for niche SPIA products) are priced more conservatively because they have a smaller population sample.

So, those offering the better deals have either taken more risk with their investments (or been able to secure access to "better" investments), with their mortality statistics (or have access to a broader population range), or are minimising their fees (either through productivity enhancements or by taking a reduction in profits).

cheers
StillGoing
Another aspect with mortality statistics is hedging with a life insurance book of business. A company with only annuities is at a much bigger risk of positive trends affecting mortality than one that can keep annuity liabilities and life insurance liabilities in balance. That way, anything that causes high mortality (say a pandemic) can trigger some life insurance claims and also extinguish some annuity obligations. Hence, if I were an insurance company, I would start pricing annuities a bit richer if I had too many relative to life insurance, etc. etc.
Topic Author
petulant
Posts: 3601
Joined: Thu Sep 22, 2016 1:09 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by petulant »

StillGoing wrote: Sun Jun 04, 2023 4:03 am
Lawrence of Suburbia wrote: Sat Jun 03, 2023 10:46 pm
Taylor Larimore wrote: Sat Jun 03, 2023 11:56 am Bogleheads:

One of the nice things about buying SPIAs (Single Premium Immediate Annuities) is that they are nearly identical (you pay a one-time premium and they guarantee a life income--like a pension or social security). Because they are nearly identical their premiums are very competitive. In my view, there is not much more to evaluate. I own two and they are the best investments we ever made.

Best wishes.
Taylor
[...]
This, I think.

Apart from caculating how the SPIA fits in with one's overall financial plan, of course. It's plainly not right for some people -- those who plainly don't have enough saved to retire, simply can't afford one; and those whose WR is less than 2% (say) just don't need one, probably.

When my monthly SPIA check hits my bank account and goes towards my bills, I don't worry about how much is my principal being returned, how much is from bonds held by the insurance company, or (well off in the future
in my case) how much is mortality credits.
I agree, from a retiree point of view how an annuity might fit into a retirement plan is the question rather than what is under the hood or value for money and the comparison should probably be with the alternatives which are (leaving aside social security/state pension and DB pensions)

1) Withdrawals from a stock/bond portfolio
2) Payments from a non-rolling bond ladder

The comparison should be appropriate too. For example, in the UK (I happen to have some recent figures available - the quotes were generated today) SPIAs are currently available such that (all for 65 yo with no guarantee period and quoted to nearest 10 basis points)

Code: Select all

Single life		Level	7.1%
			RPI	4.5%
Joint (50%) 		Level	6.6%
			RPI	4.0%
Joint (100%)		Level	6.1%
                   	RPI 	3.6%
While the SAFEMAX (in %) for a stock and bond portfolio over a 35 year period (i.e. to 100yo for a 65yo retiree). All assume 50% bonds and 50% cash in the fixed income component and use asset returns and cpi/rpi from macrohistory.net. For the level case, the nominal payout rate is fixed.

Code: Select all

		Stock allocation (%)
		40	60	80
Level		4.6	4.8	5.0
RPI		2.3	2.7	2.9
And constructing a 35 year ladder from UK Government bonds gives ladder payout rates (in %) of (I've used 15 year spot rates for 1 June 2023 downloaded from https://www.bankofengland.co.uk/statistics/yield-curves)

Code: Select all

Level	5.4
RPI	3.3
Costs for ladder and portfolio are ignored as are taxes.

In each case (level or RPI), the annuity (even joint with 100% survivor benefits) currently gives a better income rate than either ladder or portfolio (this would not have necessarily been the case 2 year ago - e.g. in July 2021, the payouts on single life annuities were 5.0% and 2.8% for level and RPI, respectively - much less attractive compared to the portfolio SAFEMAX).

However, legacy motives (in the event of early death) are a consideration as well as how well other sources of guaranteed income cover spending.

My apologies for a longish post probably only obliquely related to the topic.

cheers
StillGoing
It's amazing that the UK has such a robust market for inflation securities and annuities.
StillGoing
Posts: 359
Joined: Mon Nov 04, 2019 3:43 am
Location: U.K.

Re: Evaluating SPIAs by implied return or life expectancy

Post by StillGoing »

petulant wrote: Sun Jun 04, 2023 8:16 pm
StillGoing wrote: Sun Jun 04, 2023 1:55 am As someone who has poked under the bonnet of annuity payout rates, having attempted to match historical payout rates in both the US and UK in order to generate a time series of payout rates using historical interest rates and modern mortality statistics for backtesting purposes, this is an interesting thread. As already identified, there are 3 factors in determining the payout rate:

1) The return on investments
2) Mortality statistics
3) Fees

That different insurance companies end up using different values for each of these is exemplified by the variation in quoted prices, e.g. in the Summer 2021 edition of annuity shopper (https://www.immediateannuities.com/pdfs ... .pdf?arx=d) (this no longer seems to be updated - does anyone know of an equivalent where many quotes are given rather than just the average or maximum?), the average payout rate for a level annuity for a 65 year old male (life only) was $483 per month (per $100k premium). However, the minimum quote was $458 and the maximum quote was $503, a range of roughly plus/minus 5% of the average value. Historically, the range has been as large as +/-10% (e.g., see Figure 1 in Warshawsky, Private Annuity Markets in the United States: 1919-1984 - there are some nice calculations of load factors in that paper too).

The range of investments (UK based and for the rather specialised situation of pension buy outs) is alluded to in https://www.wtwco.com/en-gb/insights/20 ... der-invest - particularly interesting is the illiquidity premium gained on some of the investments and the emphasis on cash flow.

Mortality statistics are fairly well known (but there are many versions - the ones used in the UK by insurance companies are not publicly available), but presumably small providers (or for niche SPIA products) are priced more conservatively because they have a smaller population sample.

So, those offering the better deals have either taken more risk with their investments (or been able to secure access to "better" investments), with their mortality statistics (or have access to a broader population range), or are minimising their fees (either through productivity enhancements or by taking a reduction in profits).

cheers
StillGoing
Another aspect with mortality statistics is hedging with a life insurance book of business. A company with only annuities is at a much bigger risk of positive trends affecting mortality than one that can keep annuity liabilities and life insurance liabilities in balance. That way, anything that causes high mortality (say a pandemic) can trigger some life insurance claims and also extinguish some annuity obligations. Hence, if I were an insurance company, I would start pricing annuities a bit richer if I had too many relative to life insurance, etc. etc.
You're right this is also a consideration - it is worth noting there are more life insurance policies taken out per year (about 2m in the UK, see https://www.abi.org.uk/globalassets/fil ... s_2021.pdf) compared to annuities (about 70k in 2021/22, older data at https://www.pensionspolicyinstitute.org ... -final.pdf, but also in the previous linked document), so there is plenty of scope! I assume similar figures will apply to the US.

I also found a rather interesting paper by the UK Financial Conduct Authority (see https://www.fca.org.uk/publication/occa ... aper-5.pdf) comparing the value of money from annuities with other forms of drawdown (i.e., withdrawals from a portfolio). The paper is nearly 10 years old and 95 pages, but may be useful reading. I'll quote a short section that is relevant to your comment:
If reliable data were available on the size and components of the costs faced by annuity providers, one could assess how much of the gap between calculated MW and 100% could be attributed to these costs and how much to other factors. In practice, however, it is particularly difficult to estimate the magnitude of these costs. First, nearly all annuity providers are active in a number of other market segments (e.g. life insurance, saving products, etc.) and apportioning the costs relevant to the annuity business is no trivial job. Furthermore, an annuity provider may have a different cost of capital from its competitors because of its size and specific asset profiles.
cheers
StillGoing
StillGoing
Posts: 359
Joined: Mon Nov 04, 2019 3:43 am
Location: U.K.

Re: Evaluating SPIAs by implied return or life expectancy

Post by StillGoing »

petulant wrote: Sun Jun 04, 2023 8:19 pm
StillGoing wrote: Sun Jun 04, 2023 4:03 am
Lawrence of Suburbia wrote: Sat Jun 03, 2023 10:46 pm
Taylor Larimore wrote: Sat Jun 03, 2023 11:56 am Bogleheads:

One of the nice things about buying SPIAs (Single Premium Immediate Annuities) is that they are nearly identical (you pay a one-time premium and they guarantee a life income--like a pension or social security). Because they are nearly identical their premiums are very competitive. In my view, there is not much more to evaluate. I own two and they are the best investments we ever made.

Best wishes.
Taylor
[...]
This, I think.

Apart from caculating how the SPIA fits in with one's overall financial plan, of course. It's plainly not right for some people -- those who plainly don't have enough saved to retire, simply can't afford one; and those whose WR is less than 2% (say) just don't need one, probably.

When my monthly SPIA check hits my bank account and goes towards my bills, I don't worry about how much is my principal being returned, how much is from bonds held by the insurance company, or (well off in the future
in my case) how much is mortality credits.
I agree, from a retiree point of view how an annuity might fit into a retirement plan is the question rather than what is under the hood or value for money and the comparison should probably be with the alternatives which are (leaving aside social security/state pension and DB pensions)

1) Withdrawals from a stock/bond portfolio
2) Payments from a non-rolling bond ladder

The comparison should be appropriate too. For example, in the UK (I happen to have some recent figures available - the quotes were generated today) SPIAs are currently available such that (all for 65 yo with no guarantee period and quoted to nearest 10 basis points)

Code: Select all

Single life		Level	7.1%
			RPI	4.5%
Joint (50%) 		Level	6.6%
			RPI	4.0%
Joint (100%)		Level	6.1%
                   	RPI 	3.6%
While the SAFEMAX (in %) for a stock and bond portfolio over a 35 year period (i.e. to 100yo for a 65yo retiree). All assume 50% bonds and 50% cash in the fixed income component and use asset returns and cpi/rpi from macrohistory.net. For the level case, the nominal payout rate is fixed.

Code: Select all

		Stock allocation (%)
		40	60	80
Level		4.6	4.8	5.0
RPI		2.3	2.7	2.9
And constructing a 35 year ladder from UK Government bonds gives ladder payout rates (in %) of (I've used 15 year spot rates for 1 June 2023 downloaded from https://www.bankofengland.co.uk/statistics/yield-curves)

Code: Select all

Level	5.4
RPI	3.3
Costs for ladder and portfolio are ignored as are taxes.

In each case (level or RPI), the annuity (even joint with 100% survivor benefits) currently gives a better income rate than either ladder or portfolio (this would not have necessarily been the case 2 year ago - e.g. in July 2021, the payouts on single life annuities were 5.0% and 2.8% for level and RPI, respectively - much less attractive compared to the portfolio SAFEMAX).

However, legacy motives (in the event of early death) are a consideration as well as how well other sources of guaranteed income cover spending.

My apologies for a longish post probably only obliquely related to the topic.

cheers
StillGoing
It's amazing that the UK has such a robust market for inflation securities and annuities.
I'm not sure how many people buy inflation linked annuities in the UK, it used to be relatively small (I once saw it compared to the number of high end luxury cars of a particular make sold in the UK - but cannot now find the article), but recent high and, so far, sticky inflation has probably concentrated minds (there are currently articles in the press saying how the numbers of index linked annuities has increased, but I haven't found any firm numbers and, quite rightly in my view, an increasing amount of noise around them - although few seem to do a rigorous comparison).

About 25% of UK debt is in inflation linked securities some which has extremely long maturities (up to 50 years - the longest one really does mature in 2073! There are others that mature in 2062, 2065, and 2068 - all with real YTM about 0.8 to 1.0%). They are currently linked to RPI (although RPI will become CPI after 2030). While we are not allowed to speculate here, this worries me.

cheers
StillGoing
Topic Author
petulant
Posts: 3601
Joined: Thu Sep 22, 2016 1:09 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by petulant »

StillGoing wrote: Mon Jun 05, 2023 3:29 am
petulant wrote: Sun Jun 04, 2023 8:16 pm
StillGoing wrote: Sun Jun 04, 2023 1:55 am As someone who has poked under the bonnet of annuity payout rates, having attempted to match historical payout rates in both the US and UK in order to generate a time series of payout rates using historical interest rates and modern mortality statistics for backtesting purposes, this is an interesting thread. As already identified, there are 3 factors in determining the payout rate:

1) The return on investments
2) Mortality statistics
3) Fees

That different insurance companies end up using different values for each of these is exemplified by the variation in quoted prices, e.g. in the Summer 2021 edition of annuity shopper (https://www.immediateannuities.com/pdfs ... .pdf?arx=d) (this no longer seems to be updated - does anyone know of an equivalent where many quotes are given rather than just the average or maximum?), the average payout rate for a level annuity for a 65 year old male (life only) was $483 per month (per $100k premium). However, the minimum quote was $458 and the maximum quote was $503, a range of roughly plus/minus 5% of the average value. Historically, the range has been as large as +/-10% (e.g., see Figure 1 in Warshawsky, Private Annuity Markets in the United States: 1919-1984 - there are some nice calculations of load factors in that paper too).

The range of investments (UK based and for the rather specialised situation of pension buy outs) is alluded to in https://www.wtwco.com/en-gb/insights/20 ... der-invest - particularly interesting is the illiquidity premium gained on some of the investments and the emphasis on cash flow.

Mortality statistics are fairly well known (but there are many versions - the ones used in the UK by insurance companies are not publicly available), but presumably small providers (or for niche SPIA products) are priced more conservatively because they have a smaller population sample.

So, those offering the better deals have either taken more risk with their investments (or been able to secure access to "better" investments), with their mortality statistics (or have access to a broader population range), or are minimising their fees (either through productivity enhancements or by taking a reduction in profits).

cheers
StillGoing
Another aspect with mortality statistics is hedging with a life insurance book of business. A company with only annuities is at a much bigger risk of positive trends affecting mortality than one that can keep annuity liabilities and life insurance liabilities in balance. That way, anything that causes high mortality (say a pandemic) can trigger some life insurance claims and also extinguish some annuity obligations. Hence, if I were an insurance company, I would start pricing annuities a bit richer if I had too many relative to life insurance, etc. etc.
You're right this is also a consideration - it is worth noting there are more life insurance policies taken out per year (about 2m in the UK, see https://www.abi.org.uk/globalassets/fil ... s_2021.pdf) compared to annuities (about 70k in 2021/22, older data at https://www.pensionspolicyinstitute.org ... -final.pdf, but also in the previous linked document), so there is plenty of scope! I assume similar figures will apply to the US.

I also found a rather interesting paper by the UK Financial Conduct Authority (see https://www.fca.org.uk/publication/occa ... aper-5.pdf) comparing the value of money from annuities with other forms of drawdown (i.e., withdrawals from a portfolio). The paper is nearly 10 years old and 95 pages, but may be useful reading. I'll quote a short section that is relevant to your comment:
If reliable data were available on the size and components of the costs faced by annuity providers, one could assess how much of the gap between calculated MW and 100% could be attributed to these costs and how much to other factors. In practice, however, it is particularly difficult to estimate the magnitude of these costs. First, nearly all annuity providers are active in a number of other market segments (e.g. life insurance, saving products, etc.) and apportioning the costs relevant to the annuity business is no trivial job. Furthermore, an annuity provider may have a different cost of capital from its competitors because of its size and specific asset profiles.
cheers
StillGoing
Interesting comments. One issue i could see with just comparing annuity contracts to life insurance policies is that, I imagine, a large number of the life insurance policies are term, which would mean they would run out over a different part of most individuals' lifespans. I infer that it would take a large book of term life insurance on people in their 40s to get the right correlation to annuities for people in their 70s. I suspect permanent life insurance is rather more of a holy grail for hedging an annuity book of business.
Valuethinker
Posts: 48944
Joined: Fri May 11, 2007 11:07 am

Re: Evaluating SPIAs by implied return or life expectancy

Post by Valuethinker »

petulant wrote: Sun Jun 04, 2023 8:19 pm

It's amazing that the UK has such a robust market for inflation securities and annuities.
1. the UK had very high inflation (over 20% pa) through much of the 1970s. Inflation was generally higher than the rest of the world in the 1980s as well (although falling). Now, we have higher inflation than Europe - in part due to Brexit. Investors have, wisely, demanded an inflation premium from the UK govt.

2. About 1/4 of UK govt debt is made up of Index-Linked Gilts (ie the TIPS equivalent). It's thus possible for an insurer to hedge an inflation-linked liability. There is a (fairly small) market for IL corporate bonds as well (typically issued by the likes of utility companies, whose revenues are linked to inflation).

I don't know what the actual percentage of retirees choosing IL annuities v straight annuities. My sense is it was about 20-25%. The attractions of a 2x higher starting income (or more) were very strong and most of us don't have a good intuitive feel for the LR effects of inflation.

For UK defined benefit pension holders, by law they are indexed to inflation - up to 5% I believe.

UK state pension benefits are much smaller than Social Security, so the degree of exposure of retirees to inflation is greater. That's a problem that will become more and more apparent as the swing in numbers of retirees w Defined Contribution pensions becomes manifest (we are just about at the breakpoint where, if you are over 60 now, you joined the workforce as or before DB pensions in the private sector started to be terminated).

Also the requirement to buy an annuity with 75% of your DC balance was dropped. So strategies are now being used that allow variable withdrawals. Again we are just beginning to see consequences of that in terms of people who have spent their retirement savings (often to do things like pay off a mortgage, so there would still be housing equity there).
Topic Author
petulant
Posts: 3601
Joined: Thu Sep 22, 2016 1:09 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by petulant »

Valuethinker wrote: Tue Jun 06, 2023 5:46 am
petulant wrote: Sun Jun 04, 2023 8:19 pm

It's amazing that the UK has such a robust market for inflation securities and annuities.
1. the UK had very high inflation (over 20% pa) through much of the 1970s. Inflation was generally higher than the rest of the world in the 1980s as well (although falling). Now, we have higher inflation than Europe - in part due to Brexit. Investors have, wisely, demanded an inflation premium from the UK govt.

2. About 1/4 of UK govt debt is made up of Index-Linked Gilts (ie the TIPS equivalent). It's thus possible for an insurer to hedge an inflation-linked liability. There is a (fairly small) market for IL corporate bonds as well (typically issued by the likes of utility companies, whose revenues are linked to inflation).

I don't know what the actual percentage of retirees choosing IL annuities v straight annuities. My sense is it was about 20-25%. The attractions of a 2x higher starting income (or more) were very strong and most of us don't have a good intuitive feel for the LR effects of inflation.

For UK defined benefit pension holders, by law they are indexed to inflation - up to 5% I believe.

UK state pension benefits are much smaller than Social Security, so the degree of exposure of retirees to inflation is greater. That's a problem that will become more and more apparent as the swing in numbers of retirees w Defined Contribution pensions becomes manifest (we are just about at the breakpoint where, if you are over 60 now, you joined the workforce as or before DB pensions in the private sector started to be terminated).

Also the requirement to buy an annuity with 75% of your DC balance was dropped. So strategies are now being used that allow variable withdrawals. Again we are just beginning to see consequences of that in terms of people who have spent their retirement savings (often to do things like pay off a mortgage, so there would still be housing equity there).
Very interesting! Even 25% is pretty good relative to the United States--do you think that kind of market share will stick around or do you think it will decline with some of the transitions you mentioned?
Rex66
Posts: 2926
Joined: Tue Aug 04, 2020 5:13 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by Rex66 »

petulant wrote: Mon Jun 05, 2023 4:52 am
StillGoing wrote: Mon Jun 05, 2023 3:29 am
petulant wrote: Sun Jun 04, 2023 8:16 pm
StillGoing wrote: Sun Jun 04, 2023 1:55 am As someone who has poked under the bonnet of annuity payout rates, having attempted to match historical payout rates in both the US and UK in order to generate a time series of payout rates using historical interest rates and modern mortality statistics for backtesting purposes, this is an interesting thread. As already identified, there are 3 factors in determining the payout rate:

1) The return on investments
2) Mortality statistics
3) Fees

That different insurance companies end up using different values for each of these is exemplified by the variation in quoted prices, e.g. in the Summer 2021 edition of annuity shopper (https://www.immediateannuities.com/pdfs ... .pdf?arx=d) (this no longer seems to be updated - does anyone know of an equivalent where many quotes are given rather than just the average or maximum?), the average payout rate for a level annuity for a 65 year old male (life only) was $483 per month (per $100k premium). However, the minimum quote was $458 and the maximum quote was $503, a range of roughly plus/minus 5% of the average value. Historically, the range has been as large as +/-10% (e.g., see Figure 1 in Warshawsky, Private Annuity Markets in the United States: 1919-1984 - there are some nice calculations of load factors in that paper too).

The range of investments (UK based and for the rather specialised situation of pension buy outs) is alluded to in https://www.wtwco.com/en-gb/insights/20 ... der-invest - particularly interesting is the illiquidity premium gained on some of the investments and the emphasis on cash flow.

Mortality statistics are fairly well known (but there are many versions - the ones used in the UK by insurance companies are not publicly available), but presumably small providers (or for niche SPIA products) are priced more conservatively because they have a smaller population sample.

So, those offering the better deals have either taken more risk with their investments (or been able to secure access to "better" investments), with their mortality statistics (or have access to a broader population range), or are minimising their fees (either through productivity enhancements or by taking a reduction in profits).

cheers
StillGoing
Another aspect with mortality statistics is hedging with a life insurance book of business. A company with only annuities is at a much bigger risk of positive trends affecting mortality than one that can keep annuity liabilities and life insurance liabilities in balance. That way, anything that causes high mortality (say a pandemic) can trigger some life insurance claims and also extinguish some annuity obligations. Hence, if I were an insurance company, I would start pricing annuities a bit richer if I had too many relative to life insurance, etc. etc.
You're right this is also a consideration - it is worth noting there are more life insurance policies taken out per year (about 2m in the UK, see https://www.abi.org.uk/globalassets/fil ... s_2021.pdf) compared to annuities (about 70k in 2021/22, older data at https://www.pensionspolicyinstitute.org ... -final.pdf, but also in the previous linked document), so there is plenty of scope! I assume similar figures will apply to the US.

I also found a rather interesting paper by the UK Financial Conduct Authority (see https://www.fca.org.uk/publication/occa ... aper-5.pdf) comparing the value of money from annuities with other forms of drawdown (i.e., withdrawals from a portfolio). The paper is nearly 10 years old and 95 pages, but may be useful reading. I'll quote a short section that is relevant to your comment:
If reliable data were available on the size and components of the costs faced by annuity providers, one could assess how much of the gap between calculated MW and 100% could be attributed to these costs and how much to other factors. In practice, however, it is particularly difficult to estimate the magnitude of these costs. First, nearly all annuity providers are active in a number of other market segments (e.g. life insurance, saving products, etc.) and apportioning the costs relevant to the annuity business is no trivial job. Furthermore, an annuity provider may have a different cost of capital from its competitors because of its size and specific asset profiles.
cheers
StillGoing
Interesting comments. One issue i could see with just comparing annuity contracts to life insurance policies is that, I imagine, a large number of the life insurance policies are term, which would mean they would run out over a different part of most individuals' lifespans. I infer that it would take a large book of term life insurance on people in their 40s to get the right correlation to annuities for people in their 70s. I suspect permanent life insurance is rather more of a holy grail for hedging an annuity book of business.
even out of the permanent insurance contracts, only 15% are kept in force until death. Now that isnt a negative for the insurance company at all but it makes it hard to know how its hedged appropriately. The vast majority are lapsed early on as well.
Topic Author
petulant
Posts: 3601
Joined: Thu Sep 22, 2016 1:09 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by petulant »

Rex66 wrote: Tue Jun 06, 2023 7:31 pm
petulant wrote: Mon Jun 05, 2023 4:52 am
StillGoing wrote: Mon Jun 05, 2023 3:29 am
petulant wrote: Sun Jun 04, 2023 8:16 pm
StillGoing wrote: Sun Jun 04, 2023 1:55 am As someone who has poked under the bonnet of annuity payout rates, having attempted to match historical payout rates in both the US and UK in order to generate a time series of payout rates using historical interest rates and modern mortality statistics for backtesting purposes, this is an interesting thread. As already identified, there are 3 factors in determining the payout rate:

1) The return on investments
2) Mortality statistics
3) Fees

That different insurance companies end up using different values for each of these is exemplified by the variation in quoted prices, e.g. in the Summer 2021 edition of annuity shopper (https://www.immediateannuities.com/pdfs ... .pdf?arx=d) (this no longer seems to be updated - does anyone know of an equivalent where many quotes are given rather than just the average or maximum?), the average payout rate for a level annuity for a 65 year old male (life only) was $483 per month (per $100k premium). However, the minimum quote was $458 and the maximum quote was $503, a range of roughly plus/minus 5% of the average value. Historically, the range has been as large as +/-10% (e.g., see Figure 1 in Warshawsky, Private Annuity Markets in the United States: 1919-1984 - there are some nice calculations of load factors in that paper too).

The range of investments (UK based and for the rather specialised situation of pension buy outs) is alluded to in https://www.wtwco.com/en-gb/insights/20 ... der-invest - particularly interesting is the illiquidity premium gained on some of the investments and the emphasis on cash flow.

Mortality statistics are fairly well known (but there are many versions - the ones used in the UK by insurance companies are not publicly available), but presumably small providers (or for niche SPIA products) are priced more conservatively because they have a smaller population sample.

So, those offering the better deals have either taken more risk with their investments (or been able to secure access to "better" investments), with their mortality statistics (or have access to a broader population range), or are minimising their fees (either through productivity enhancements or by taking a reduction in profits).

cheers
StillGoing
Another aspect with mortality statistics is hedging with a life insurance book of business. A company with only annuities is at a much bigger risk of positive trends affecting mortality than one that can keep annuity liabilities and life insurance liabilities in balance. That way, anything that causes high mortality (say a pandemic) can trigger some life insurance claims and also extinguish some annuity obligations. Hence, if I were an insurance company, I would start pricing annuities a bit richer if I had too many relative to life insurance, etc. etc.
You're right this is also a consideration - it is worth noting there are more life insurance policies taken out per year (about 2m in the UK, see https://www.abi.org.uk/globalassets/fil ... s_2021.pdf) compared to annuities (about 70k in 2021/22, older data at https://www.pensionspolicyinstitute.org ... -final.pdf, but also in the previous linked document), so there is plenty of scope! I assume similar figures will apply to the US.

I also found a rather interesting paper by the UK Financial Conduct Authority (see https://www.fca.org.uk/publication/occa ... aper-5.pdf) comparing the value of money from annuities with other forms of drawdown (i.e., withdrawals from a portfolio). The paper is nearly 10 years old and 95 pages, but may be useful reading. I'll quote a short section that is relevant to your comment:
If reliable data were available on the size and components of the costs faced by annuity providers, one could assess how much of the gap between calculated MW and 100% could be attributed to these costs and how much to other factors. In practice, however, it is particularly difficult to estimate the magnitude of these costs. First, nearly all annuity providers are active in a number of other market segments (e.g. life insurance, saving products, etc.) and apportioning the costs relevant to the annuity business is no trivial job. Furthermore, an annuity provider may have a different cost of capital from its competitors because of its size and specific asset profiles.
cheers
StillGoing
Interesting comments. One issue i could see with just comparing annuity contracts to life insurance policies is that, I imagine, a large number of the life insurance policies are term, which would mean they would run out over a different part of most individuals' lifespans. I infer that it would take a large book of term life insurance on people in their 40s to get the right correlation to annuities for people in their 70s. I suspect permanent life insurance is rather more of a holy grail for hedging an annuity book of business.
even out of the permanent insurance contracts, only 15% are kept in force until death. Now that isnt a negative for the insurance company at all but it makes it hard to know how its hedged appropriately. The vast majority are lapsed early on as well.
I imagine the policies that are in force with insureds in their 70s are known to them for hedging purposes. I also imagine that they can derive some kind of mortality correlation between all other policies on younger period (perhaps subdivided into blocs) based on some kind of "beta" with older populations typically annuitizing.
Valuethinker
Posts: 48944
Joined: Fri May 11, 2007 11:07 am

Re: Evaluating SPIAs by implied return or life expectancy

Post by Valuethinker »

petulant wrote: Tue Jun 06, 2023 6:26 pm
Valuethinker wrote: Tue Jun 06, 2023 5:46 am
petulant wrote: Sun Jun 04, 2023 8:19 pm

It's amazing that the UK has such a robust market for inflation securities and annuities.
1. the UK had very high inflation (over 20% pa) through much of the 1970s. Inflation was generally higher than the rest of the world in the 1980s as well (although falling). Now, we have higher inflation than Europe - in part due to Brexit. Investors have, wisely, demanded an inflation premium from the UK govt.

2. About 1/4 of UK govt debt is made up of Index-Linked Gilts (ie the TIPS equivalent). It's thus possible for an insurer to hedge an inflation-linked liability. There is a (fairly small) market for IL corporate bonds as well (typically issued by the likes of utility companies, whose revenues are linked to inflation).

I don't know what the actual percentage of retirees choosing IL annuities v straight annuities. My sense is it was about 20-25%. The attractions of a 2x higher starting income (or more) were very strong and most of us don't have a good intuitive feel for the LR effects of inflation.

For UK defined benefit pension holders, by law they are indexed to inflation - up to 5% I believe.

UK state pension benefits are much smaller than Social Security, so the degree of exposure of retirees to inflation is greater. That's a problem that will become more and more apparent as the swing in numbers of retirees w Defined Contribution pensions becomes manifest (we are just about at the breakpoint where, if you are over 60 now, you joined the workforce as or before DB pensions in the private sector started to be terminated).

Also the requirement to buy an annuity with 75% of your DC balance was dropped. So strategies are now being used that allow variable withdrawals. Again we are just beginning to see consequences of that in terms of people who have spent their retirement savings (often to do things like pay off a mortgage, so there would still be housing equity there).
Very interesting! Even 25% is pretty good relative to the United States--do you think that kind of market share will stick around or do you think it will decline with some of the transitions you mentioned?
It's government funding policy.

The ILG proportion was:
- meant to show commitment to future low inflation
- set to match the requirements of investors and in particular pension funds & annuity providers

Of course it has turned out to be relatively expensive funding for the government. Had they borrowed fixed at the rates available the last few years (2.0% or even lower) then there would have been a huge gain arising from 10% inflation. Instead, the "problem" for the Chancellor (Min of Finance) is the cost of borrowing has risen with inflation -- due to the accretion of the coupons and principal on the ILG.

So the borrowing strategy worked in that it protected investors and in particular pensioners against a large burst of unexpected inflation. (Prior to 2022, ILGs were being issued at c -2.5% real yield on the 20 year ILG, so the government also was borrowing "cheaply" (ie a negative real rate) except, due to that 10% inflation, it has turned out to be more expensive than planned).

Probably HM Treasury decides this is expensive funding and reduces the future stake. Canada is abandoning the issuance of Real Return Bonds. This more than anything gives me some pause as to the future track of inflation - a credible commitment to keep it low is being slowly jettisoned.

UK govt debt funding policy (the Debt Management Office or DMO) certainly bears watching. I imagine the Chancellor is making some fairly strident views known there.

(A similar study for US Treasury, before this inflationary period, showed TIPS had been an expensive form of borrowing. However that was largely because of high real yields in the early years post their issuance - the late 1990s and early 2000s. So I hope that gets disregarded and the product remains).

The inflation-linked corporate bond market is quite small. Because of the importance of the inflation accretion element to principal, the returns are "back weighted" in the bond's life, and that means the credit risk is higher (for any given credit rating).
Rex66
Posts: 2926
Joined: Tue Aug 04, 2020 5:13 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by Rex66 »

petulant wrote: Tue Jun 06, 2023 8:20 pm
Rex66 wrote: Tue Jun 06, 2023 7:31 pm
petulant wrote: Mon Jun 05, 2023 4:52 am
StillGoing wrote: Mon Jun 05, 2023 3:29 am
petulant wrote: Sun Jun 04, 2023 8:16 pm

Another aspect with mortality statistics is hedging with a life insurance book of business. A company with only annuities is at a much bigger risk of positive trends affecting mortality than one that can keep annuity liabilities and life insurance liabilities in balance. That way, anything that causes high mortality (say a pandemic) can trigger some life insurance claims and also extinguish some annuity obligations. Hence, if I were an insurance company, I would start pricing annuities a bit richer if I had too many relative to life insurance, etc. etc.
You're right this is also a consideration - it is worth noting there are more life insurance policies taken out per year (about 2m in the UK, see https://www.abi.org.uk/globalassets/fil ... s_2021.pdf) compared to annuities (about 70k in 2021/22, older data at https://www.pensionspolicyinstitute.org ... -final.pdf, but also in the previous linked document), so there is plenty of scope! I assume similar figures will apply to the US.

I also found a rather interesting paper by the UK Financial Conduct Authority (see https://www.fca.org.uk/publication/occa ... aper-5.pdf) comparing the value of money from annuities with other forms of drawdown (i.e., withdrawals from a portfolio). The paper is nearly 10 years old and 95 pages, but may be useful reading. I'll quote a short section that is relevant to your comment:
If reliable data were available on the size and components of the costs faced by annuity providers, one could assess how much of the gap between calculated MW and 100% could be attributed to these costs and how much to other factors. In practice, however, it is particularly difficult to estimate the magnitude of these costs. First, nearly all annuity providers are active in a number of other market segments (e.g. life insurance, saving products, etc.) and apportioning the costs relevant to the annuity business is no trivial job. Furthermore, an annuity provider may have a different cost of capital from its competitors because of its size and specific asset profiles.
cheers
StillGoing
Interesting comments. One issue i could see with just comparing annuity contracts to life insurance policies is that, I imagine, a large number of the life insurance policies are term, which would mean they would run out over a different part of most individuals' lifespans. I infer that it would take a large book of term life insurance on people in their 40s to get the right correlation to annuities for people in their 70s. I suspect permanent life insurance is rather more of a holy grail for hedging an annuity book of business.
even out of the permanent insurance contracts, only 15% are kept in force until death. Now that isnt a negative for the insurance company at all but it makes it hard to know how its hedged appropriately. The vast majority are lapsed early on as well.
I imagine the policies that are in force with insureds in their 70s are known to them for hedging purposes. I also imagine that they can derive some kind of mortality correlation between all other policies on younger period (perhaps subdivided into blocs) based on some kind of "beta" with older populations typically annuitizing.
Nobody really knows but older typically shouldn’t annuitize but instead loan. Most of the permanent companies aren’t the same companies that you might buy annuities. Much of the benefits or difficulties seem to get incorporated in the dividends.
I’m sure there is something they get out of it but I doubt it’s a full hedge.
StillGoing
Posts: 359
Joined: Mon Nov 04, 2019 3:43 am
Location: U.K.

Re: Evaluating SPIAs by implied return or life expectancy

Post by StillGoing »

Valuethinker wrote: Tue Jun 06, 2023 5:46 am
petulant wrote: Sun Jun 04, 2023 8:19 pm

It's amazing that the UK has such a robust market for inflation securities and annuities.
1. the UK had very high inflation (over 20% pa) through much of the 1970s. Inflation was generally higher than the rest of the world in the 1980s as well (although falling). Now, we have higher inflation than Europe - in part due to Brexit. Investors have, wisely, demanded an inflation premium from the UK govt.

2. About 1/4 of UK govt debt is made up of Index-Linked Gilts (ie the TIPS equivalent). It's thus possible for an insurer to hedge an inflation-linked liability. There is a (fairly small) market for IL corporate bonds as well (typically issued by the likes of utility companies, whose revenues are linked to inflation).

I don't know what the actual percentage of retirees choosing IL annuities v straight annuities. My sense is it was about 20-25%. The attractions of a 2x higher starting income (or more) were very strong and most of us don't have a good intuitive feel for the LR effects of inflation.

For UK defined benefit pension holders, by law they are indexed to inflation - up to 5% I believe.

UK state pension benefits are much smaller than Social Security, so the degree of exposure of retirees to inflation is greater. That's a problem that will become more and more apparent as the swing in numbers of retirees w Defined Contribution pensions becomes manifest (we are just about at the breakpoint where, if you are over 60 now, you joined the workforce as or before DB pensions in the private sector started to be terminated).

Also the requirement to buy an annuity with 75% of your DC balance was dropped. So strategies are now being used that allow variable withdrawals. Again we are just beginning to see consequences of that in terms of people who have spent their retirement savings (often to do things like pay off a mortgage, so there would still be housing equity there).
The number of escalation or level annuities can be found in the full data tables at https://www.fca.org.uk/data/retirement- ... 21-22#full

From October 2021 to March 2022, it appears that level annuities were 87% of those sold in the UK and escalating annuities (i.e. those with fixed COLA or fully index linked, there isn't a breakdown between those types) were 13%. These proportions were fairly constant over the period for which data are available (2015-2022). It will be interesting to see the next release (I think in October 2023) covering the period where worries about inflation and increases in payout rates have occurred. Certainly on some of the UK based boards I frequent, there appears to be general move towards inflation linked annuities (at least for longevity income protection).

BTW, that spreadsheet also contains some other interesting information - e.g. 75% of annuities sold were single life (which surprised me).

For DB pensions, prior to 1998(?) there was no requirement for inflation linking (so quite a lot of benefits accumulated prior to 1998 were level). After that date until 2006(IIRC), there was a legal requirement to provide at least a 5% cap (this is sometimes known as the limited price index, LPI), while after 2006, the cap was reduced to 2.5% where it still resides. You can actually buy annuities with LPI protection that have a payout rate about 10 basis points higher than full RPI protection (apparently they are quite difficult for actuaries to price reliably) - I hope no-one buys them!

I would echo valuethinker's comments about retirement being quite different in the UK. As he says, our state pension (i.e., social security equivalent) is relatively small (£10.5k per year in 2023/24). But, we continue to largely have free at point of service medical care while care home costs are capped (both of which appear to be major concerns for US retirees).

cheers
StillGoing
User avatar
NearlyRetired
Posts: 350
Joined: Fri Aug 09, 2019 12:56 pm
Location: United Kingdom

Re: Evaluating SPIAs by implied return or life expectancy

Post by NearlyRetired »

StillGoing wrote: Wed Jun 07, 2023 7:01 am
Valuethinker wrote: Tue Jun 06, 2023 5:46 am
petulant wrote: Sun Jun 04, 2023 8:19 pm

It's amazing that the UK has such a robust market for inflation securities and annuities.
1. the UK had very high inflation (over 20% pa) through much of the 1970s. Inflation was generally higher than the rest of the world in the 1980s as well (although falling). Now, we have higher inflation than Europe - in part due to Brexit. Investors have, wisely, demanded an inflation premium from the UK govt.

2. About 1/4 of UK govt debt is made up of Index-Linked Gilts (ie the TIPS equivalent). It's thus possible for an insurer to hedge an inflation-linked liability. There is a (fairly small) market for IL corporate bonds as well (typically issued by the likes of utility companies, whose revenues are linked to inflation).

I don't know what the actual percentage of retirees choosing IL annuities v straight annuities. My sense is it was about 20-25%. The attractions of a 2x higher starting income (or more) were very strong and most of us don't have a good intuitive feel for the LR effects of inflation.

For UK defined benefit pension holders, by law they are indexed to inflation - up to 5% I believe.

UK state pension benefits are much smaller than Social Security, so the degree of exposure of retirees to inflation is greater. That's a problem that will become more and more apparent as the swing in numbers of retirees w Defined Contribution pensions becomes manifest (we are just about at the breakpoint where, if you are over 60 now, you joined the workforce as or before DB pensions in the private sector started to be terminated).

Also the requirement to buy an annuity with 75% of your DC balance was dropped. So strategies are now being used that allow variable withdrawals. Again we are just beginning to see consequences of that in terms of people who have spent their retirement savings (often to do things like pay off a mortgage, so there would still be housing equity there).
The number of escalation or level annuities can be found in the full data tables at https://www.fca.org.uk/data/retirement- ... 21-22#full

From October 2021 to March 2022, it appears that level annuities were 87% of those sold in the UK and escalating annuities (i.e. those with fixed COLA or fully index linked, there isn't a breakdown between those types) were 13%. These proportions were fairly constant over the period for which data are available (2015-2022). It will be interesting to see the next release (I think in October 2023) covering the period where worries about inflation and increases in payout rates have occurred. Certainly on some of the UK based boards I frequent, there appears to be general move towards inflation linked annuities (at least for longevity income protection).

BTW, that spreadsheet also contains some other interesting information - e.g. 75% of annuities sold were single life (which surprised me).

For DB pensions, prior to 1998(?) there was no requirement for inflation linking (so quite a lot of benefits accumulated prior to 1998 were level). After that date until 2006(IIRC), there was a legal requirement to provide at least a 5% cap (this is sometimes known as the limited price index, LPI), while after 2006, the cap was reduced to 2.5% where it still resides. You can actually buy annuities with LPI protection that have a payout rate about 10 basis points higher than full RPI protection (apparently they are quite difficult for actuaries to price reliably) - I hope no-one buys them!

I would echo valuethinker's comments about retirement being quite different in the UK. As he says, our state pension (i.e., social security equivalent) is relatively small (£10.5k per year in 2023/24). But, we continue to largely have free at point of service medical care while care home costs are capped (both of which appear to be major concerns for US retirees).

cheers
StillGoing
Without wishing to derail this thread, the other thing that struck me from the FCA data, was the fact that for the majority of pensions, people were withdrawing 8% or more of their pension pot on a regular basis (Figure 3) - which doesn't bode well for the future, and will I suspect push more people towards annuities
To err is to be human, to really mess up, use a computer
StillGoing
Posts: 359
Joined: Mon Nov 04, 2019 3:43 am
Location: U.K.

Re: Evaluating SPIAs by implied return or life expectancy

Post by StillGoing »

NearlyRetired wrote: Thu Jun 08, 2023 1:13 pm
StillGoing wrote: Wed Jun 07, 2023 7:01 am
Valuethinker wrote: Tue Jun 06, 2023 5:46 am
petulant wrote: Sun Jun 04, 2023 8:19 pm

It's amazing that the UK has such a robust market for inflation securities and annuities.
1. the UK had very high inflation (over 20% pa) through much of the 1970s. Inflation was generally higher than the rest of the world in the 1980s as well (although falling). Now, we have higher inflation than Europe - in part due to Brexit. Investors have, wisely, demanded an inflation premium from the UK govt.

2. About 1/4 of UK govt debt is made up of Index-Linked Gilts (ie the TIPS equivalent). It's thus possible for an insurer to hedge an inflation-linked liability. There is a (fairly small) market for IL corporate bonds as well (typically issued by the likes of utility companies, whose revenues are linked to inflation).

I don't know what the actual percentage of retirees choosing IL annuities v straight annuities. My sense is it was about 20-25%. The attractions of a 2x higher starting income (or more) were very strong and most of us don't have a good intuitive feel for the LR effects of inflation.

For UK defined benefit pension holders, by law they are indexed to inflation - up to 5% I believe.

UK state pension benefits are much smaller than Social Security, so the degree of exposure of retirees to inflation is greater. That's a problem that will become more and more apparent as the swing in numbers of retirees w Defined Contribution pensions becomes manifest (we are just about at the breakpoint where, if you are over 60 now, you joined the workforce as or before DB pensions in the private sector started to be terminated).

Also the requirement to buy an annuity with 75% of your DC balance was dropped. So strategies are now being used that allow variable withdrawals. Again we are just beginning to see consequences of that in terms of people who have spent their retirement savings (often to do things like pay off a mortgage, so there would still be housing equity there).
The number of escalation or level annuities can be found in the full data tables at https://www.fca.org.uk/data/retirement- ... 21-22#full

From October 2021 to March 2022, it appears that level annuities were 87% of those sold in the UK and escalating annuities (i.e. those with fixed COLA or fully index linked, there isn't a breakdown between those types) were 13%. These proportions were fairly constant over the period for which data are available (2015-2022). It will be interesting to see the next release (I think in October 2023) covering the period where worries about inflation and increases in payout rates have occurred. Certainly on some of the UK based boards I frequent, there appears to be general move towards inflation linked annuities (at least for longevity income protection).

BTW, that spreadsheet also contains some other interesting information - e.g. 75% of annuities sold were single life (which surprised me).

For DB pensions, prior to 1998(?) there was no requirement for inflation linking (so quite a lot of benefits accumulated prior to 1998 were level). After that date until 2006(IIRC), there was a legal requirement to provide at least a 5% cap (this is sometimes known as the limited price index, LPI), while after 2006, the cap was reduced to 2.5% where it still resides. You can actually buy annuities with LPI protection that have a payout rate about 10 basis points higher than full RPI protection (apparently they are quite difficult for actuaries to price reliably) - I hope no-one buys them!

I would echo valuethinker's comments about retirement being quite different in the UK. As he says, our state pension (i.e., social security equivalent) is relatively small (£10.5k per year in 2023/24). But, we continue to largely have free at point of service medical care while care home costs are capped (both of which appear to be major concerns for US retirees).

cheers
StillGoing
Without wishing to derail this thread, the other thing that struck me from the FCA data, was the fact that for the majority of pensions, people were withdrawing 8% or more of their pension pot on a regular basis (Figure 3) - which doesn't bode well for the future, and will I suspect push more people towards annuities
I've seen a few explanations for the high withdrawal rates:
1) People generally underestimate their life expectancy (my understanding is that this is one of the reasons behind the annuity puzzle - and at least relevant to this thread!)
2) That early withdrawals can be higher (in the UK) for tax reasons. For example, assuming an income below the tax threshold (e.g. particularly before state pension), then it can make sense to withdraw from the DC pension first.
3) Early withdrawals are used to pay off the mortgage and buy 'trip of a lifetime' holidays.
4) People have savings and investments elsewhere (the DC pension is still 'bedding in') not captured by the FCA or DB income (there are still over 9m people with private DB pensions and 7m with public sector pensions in the UK - there may be some overlap between those numbers). Some of the stats by UK government (https://www.gov.uk/government/statistic ... 21-to-2022) and ONS (e.g., https://www.ons.gov.uk/peoplepopulation ... nvestments) are also interesting.

Of course, 2 and 3 only explain high withdrawal rates early on in retirement.

With apologies to petulant - only 25% of the above is on topic...

cheers
StillGoing
Rex66
Posts: 2926
Joined: Tue Aug 04, 2020 5:13 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by Rex66 »

While it might be problematic, if you look at defined benefit plans which are pension plans then look at the maximum the government will allow you to take as a single lump sum at 59.5 you will find that to be around 3.4 million currently. If you look at the max it’s allowed to pay out as a yearly salary if you took that option instead it’s around 265k. That’s around 7.5% of the 3.4. I find that interesting considering the safe withdrawal rates we discuss.
Topic Author
petulant
Posts: 3601
Joined: Thu Sep 22, 2016 1:09 pm

Re: Evaluating SPIAs by implied return or life expectancy

Post by petulant »

Rex66 wrote: Fri Jun 09, 2023 5:45 am While it might be problematic, if you look at defined benefit plans which are pension plans then look at the maximum the government will allow you to take as a single lump sum at 59.5 you will find that to be around 3.4 million currently. If you look at the max it’s allowed to pay out as a yearly salary if you took that option instead it’s around 265k. That’s around 7.5% of the 3.4. I find that interesting considering the safe withdrawal rates we discuss.
Is that with no COLA or inflation adjustment? I feel like the last time I tried to figure out a non-inflation SWR it was like 6%. I would have to re-run things to be sure.
Post Reply