Composition and safety of TIAA traditional annuity

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Thu Oct 17, 2019 8:02 pm

Northern Flicker wrote:
Thu Oct 17, 2019 2:25 pm
Tail risk of TIAA Trad thus is probably slightly higher than total bond, but garden variety risk is lower because TIAA normally absorbs term risk and credit risk in return for lower liquidity.
For some perspective on this, recall that TIAA itself says, "while the 3% guarantee rate exists as a term within the TIAA Traditional Annuity contract, the term does not appear to provide any meaningful value to owners of the contract..." (cited on p. 20 of the paper). So the real value of Traditional could only be in its risk reduction. It is true that even using a 9-year-withdrawal-based measure of risk, Traditional is less risky than its duration- and quality-matched alternatives (the paper's Figure 3). It is also true, as Figure 4 shows, that Traditional's 9-year return divided by its 9-year standard deviation, which is like a Sharpe Ratio without a risk-free asset, was higher than that of any of the alternatives. But as pages 22--24 discuss, the 9-year returns were so non-normally distributed that the Sharpe Ratio is a completely misleading indicator of which investment was better, and Traditional was second-degree stochastic-dominated by several of the alternatives: the risk reduction offered by Traditional would not have been worth it to any risk-averse expected-utility-maximizing investor.

This might change in the future. But perhaps the more important value of Traditional is revealed once one takes behavioral considerations into account. Traditional's withdrawal restrictions enable it to offer a product with an almost-8-year duration which on paper never loses money, and thus which offers short-term safety. Investors who would be panicked by the daily or monthly fluctuations of a bond mutual fund with an 8-year duration can feel secure and comfortable holding Traditional's version of it. Historically, such a portfolio has returned more than money-market funds, so investors seeking the safety of an investment which never loses money have been well served by Traditional. However, an investor who understands bonds well enough to understand the irrelevance of bond funds' daily, weekly, monthly, or even yearly fluctuations would have been able to do better than Traditional while matching the duration and quality of the investments which underlie it.

(If one's fixed-income investments are undertaken for reasons alien to those considered in the paper, such as to attempt to counteract short-term movements in the equity markets, then very different considerations would enter and even the daily fluctuations of a bond fund could have importance.)

columbia
Posts: 1978
Joined: Tue Aug 27, 2013 5:30 am

Re: Composition and safety of TIAA traditional annuity

Post by columbia » Thu Oct 17, 2019 8:47 pm

I just checked and TIAA RC represents 63% of my total portfolio; if that doesn’t work out, it will not be because of TIAA, my savings rate or my equity choices. It will be because something fundamentally went wrong with global capitalism.

Trurl Klapaucius
Posts: 53
Joined: Tue Aug 06, 2013 1:11 pm

Re: Composition and safety of TIAA traditional annuity

Post by Trurl Klapaucius » Fri Oct 18, 2019 6:07 am

Gabriel A. Lozada wrote:
Thu Oct 17, 2019 12:18 am
...some of TIAA's recent public communications about Traditional report incorrect performance figures.
Given the “inconsistent” fashion TIAA Traditional crediting rates are reported, as described in the appendix of your paper, can the reported credit quality and duration of their holdings be trusted? In your study, were you able to perform an analysis similar to the one presented in the appendix to verify the consistency of reported credit quality and duration? Did this confirm that they were reported correctly?

By the way, you definitely have a future as an investigative reporter, should you decide to change jobs. :happy

dknightd
Posts: 1860
Joined: Wed Mar 07, 2018 11:57 am

Re: Composition and safety of TIAA traditional annuity

Post by dknightd » Fri Oct 18, 2019 7:49 am

Trurl Klapaucius wrote:
Wed Oct 16, 2019 5:09 pm
Returning to the original post, here is an interesting article regarding the composition and safety of the “public fixed income” component of TIAA’s investment portfolio:

Bonds for Retirement Saving: TIAA Traditional’s Lessons on Quality, Duration, Risk, and Gradual Withdrawals by Gabriel A. Lozada Department of Economics, University of Utah Salt Lake City, UT 84112, USA http://content.csbs.utah.edu/~lozada/Re ... A_4_UU.pdf

The author constructs benchmarks for comparison, mostly with Vanguard funds, and concludes:
“…risk-averse expected-utility-maximizing consumers withdrawing funds over nine years would have preferred several of the mutual-fund alternatives over Traditional had they known the probability distributions in advance.”

Is this actionable information; e.g., buy funds instead of TIAA Traditional? Or is it just another example of 20/20 vision through the retrospectoscope?
I've read parts of this paper, and parts of the authors comments in this thread.

A couple of thoughts.
The paper addresses the “public fixed income” portion of TIAA traditional. A significant part of Traditional's investments are not available to the public.
The Traditional account has considerable reserves. This allows them to guarantee, at least for some period of time, an interest rate larger than what the “public fixed income” portion of its holdings could support.

I consider TIAA Traditional to be very safe. It may, or may not, return what a mix of bond holdings might return. Remember, it is not an "investment" it is an insurance contract. So I have some TIAA traditional, and some bond funds ;)

dknightd
Posts: 1860
Joined: Wed Mar 07, 2018 11:57 am

Re: Composition and safety of TIAA traditional annuity

Post by dknightd » Fri Oct 18, 2019 8:53 am

Gabriel A. Lozada wrote:
Thu Oct 17, 2019 8:02 pm
Traditional was second-degree stochastic-dominated by several of the alternatives: the risk reduction offered by Traditional would not have been worth it to any risk-averse expected-utility-maximizing investor.
Can you explain what a "risk-averse expected-utility-maximizing investor" is?
Thanks


Edit: Also I'm curious, you work at a University so likely qualify to buy TIAA Traditional contracts, do you own any? Why, or why not?
Again, Thanks

dknightd
Posts: 1860
Joined: Wed Mar 07, 2018 11:57 am

Re: Composition and safety of TIAA traditional annuity

Post by dknightd » Fri Oct 18, 2019 9:34 am

Gabriel,
To me, from your paper, this sums it up for me:
"No education is necessary for Traditional’s investors to avoid hasty selling because they are not permitted to do it, and that very lack of permission enables Traditional to use smoothing accounting methods which calm investors so they are quite unlikely to panic in the first place. Our careful examination of Traditional has however shown that it provides no ‘free lunch.’"

TIAA Traditional was designed to provide educators with a stable and secure retirement. Without those educators spending time to get educated about investing. It was only later that TIAA offered other investment options.
Having read your entire paper now, my take is TIAA Traditional provides "no free lunch", but it also helps prevent a bully from taking your lunch money.

Full disclosure. I have both RA and SRA Traditional holdings. I plan to annuitize RA holdings. For now SRA holdings I consider a very good alternative to cash/MM/CD (it is very hard to beat safe 3% tax deferred earnings right now).

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Fri Oct 18, 2019 11:24 am

student wrote:
Thu Oct 17, 2019 7:28 pm
Gabriel A. Lozada wrote:
Thu Oct 17, 2019 2:36 pm
Such technical details aside, we've learned from Trurl Klapaucius is that TIAA's General Account quality is somewhat worse now than it was at year-end 2015.
I wonder whether this is due to the low interest rate environment. Since TIAA has contractual obligations to credit TIAA Traditional participants with at least 3% crediting rate, it has to find the yield somewhere.
TIAA has not publicly discussed changes in General Account quality, but I've found some information about general insurance industry trends.

One source a Financial Times article, "Search for yield draws US life insurers to risky places (Feb 21, 2019)" https://www.ft.com/content/7ddcd942-35c ... 2a211d90d5. This article discusses "Fitch Ratings: US Life Insurers' Investments More Risky Since Financial Crisis" (21 Feb 2019) https://www.fitchratings.com/site/pr/10062907. According to these sources, in general, insurance companies' portfolios have lower-rated debt now than before the Great Recession began. Part of the reason is because there is less high-quality corporate debt outstanding than there used to be. One of the people interviewed by the Financial Times believes that insurance industry portfolios are more diversified than they used to be, with more (taxable) municipal bonds, bank loans, private placements, and "structured products," so that "credit quality is down but default risk in portfolios is not higher."

Another source is "Positioning Life Insurance Portfolios for the Next Recession (and Beyond)" https://www.guggenheimpartners.com/pers ... -recession. Two of the charts there are particularly interesting. The first is entitled "Small Increases in Below Investment-Grade Holdings Can Put Capital at Risk"; it shows that below-investment-grade holdings were actually less on 12.31.2018 than they were in 2015. The second chart is entitled "NAIC 2 Holdings Have Surpassed Crisis Highs." The NAIC 2 rating is equivalent to BBB, and the chart shows that insurance industry holdings of it have risen since 2015.

crefwatch
Posts: 263
Joined: Sun Apr 15, 2007 1:07 pm
Location: New Jersey, USA

Re: Composition and safety of TIAA traditional annuity

Post by crefwatch » Fri Oct 18, 2019 12:05 pm

Trurl Klapaucius wrote:
Wed Oct 16, 2019 5:09 pm
Bonds for Retirement Saving: TIAA Traditional’s Lessons on Quality, Duration, Risk, and Gradual Withdrawals by Gabriel A. Lozada Department of Economics, University of Utah Salt Lake City, UT 84112, USA http://content.csbs.utah.edu/~lozada/Re ... A_4_UU.pdf
Clearly, most of the posters here know the difference, but for less sophisticated, or more casual readers of this thread, I must point out that the title of the scholarly paper is way off of "accurate." It is completely wrong to equate a consumer (investor's) view of the TIAA Traditional Annuity product with the structure of the TIAA General Account.

If you are afraid that your TIAA Traditional money will disappear, then you do care about prudent investing by this New York State regulated insurance company. But millions of people buy retirement investments from less highly-rated insurance companies every day. TIAA Traditional is a contract with the investor for TIAA to perform certain financial acts for them. It is not a Separate Account, and it is not fractional ownership of anything. It certainly depends on the General Account's success to not disappear, but it does not depend on the General Account's success to pay next year's promised benefits.

To the average lay investor, it is misleading to compare TIAA Traditional to Open-End Bond Mutual Funds. It is also misleading to think about Open-End Mutual Funds and the large amount of TIAA Traditional that is locked up for nine years and a day as if they were comparable obligations of a financial-services provider.

Between 1928 and 1941 [Greenough ... ] TIAA made multiple changes to the annuity structure to reflect better knowledge of actuarial realities. But there was a crisis in 1948 over obligations incurred (to us, that is) before 1936. Between a large grant from the Carnegie Corporation, and increasing reserves at TIAA, those investors were protected. It's wrong to assume that exact protection could happen again, especially since the company is orders-of-magnitude larger than it was in 1948. But I don't lie awake at night worrying about the General Account!

Sure, it's sensible to ask whether TIAA might go belly-up after 100 years, especially since the financial world (and the company) have changed so much since 1987. But I repeat that the title of the paper is a disservice to the scholarship.

User avatar
vineviz
Posts: 5391
Joined: Tue May 15, 2018 1:55 pm

Re: Composition and safety of TIAA traditional annuity

Post by vineviz » Fri Oct 18, 2019 12:12 pm

crefwatch wrote:
Fri Oct 18, 2019 12:05 pm
Trurl Klapaucius wrote:
Wed Oct 16, 2019 5:09 pm
Bonds for Retirement Saving: TIAA Traditional’s Lessons on Quality, Duration, Risk, and Gradual Withdrawals by Gabriel A. Lozada Department of Economics, University of Utah Salt Lake City, UT 84112, USA http://content.csbs.utah.edu/~lozada/Re ... A_4_UU.pdf
Clearly, most of the posters here know the difference, but for less sophisticated, or more casual readers of this thread, I must point out that the title of the scholarly paper is way off of "accurate." It is completely wrong to equate a consumer (investor's) view of the TIAA Traditional Annuity product with the structure of the TIAA General Account.

If you are afraid that your TIAA Traditional money will disappear, then you do care about prudent investing by this New York State regulated insurance company. But millions of people buy retirement investments from less highly-rated insurance companies every day. TIAA Traditional is a contract with the investor for TIAA to perform certain financial acts for them. It is not a Separate Account, and it is not fractional ownership of anything. It certainly depends on the General Account's success to not disappear, but it does not depend on the General Account's success to pay next year's promised benefits.

To the average lay investor, it is misleading to compare TIAA Traditional to Open-End Bond Mutual Funds. It is also misleading to think about Open-End Mutual Funds and the large amount of TIAA Traditional that is locked up for nine years and a day as if they were comparable obligations of a financial-services provider.

Between 1928 and 1941 [Greenough ... ] TIAA made multiple changes to the annuity structure to reflect better knowledge of actuarial realities. But there was a crisis in 1948 over obligations incurred (to us, that is) before 1936. Between a large grant from the Carnegie Corporation, and increasing reserves at TIAA, those investors were protected. It's wrong to assume that exact protection could happen again, especially since the company is orders-of-magnitude larger than it was in 1948. But I don't lie awake at night worrying about the General Account!

Sure, it's sensible to ask whether TIAA might go belly-up after 100 years, especially since the financial world (and the company) have changed so much since 1987. But I repeat that the title of the paper is a disservice to the scholarship.
The point of the research, which is both relevant and accurate, is to explore the conditions under which normal investors might be able to replicate the returns of TIAA “annuity”.

The finding that such replication isn’t especially difficult, at least economically, should IMHO be of interest to investors who might otherwise contort their investment plans to include TIAA (or other stable value funds, like the TSP’s G Fund).
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

student
Posts: 4133
Joined: Fri Apr 03, 2015 6:58 am

Re: Composition and safety of TIAA traditional annuity

Post by student » Fri Oct 18, 2019 12:24 pm

Gabriel A. Lozada wrote:
Fri Oct 18, 2019 11:24 am
student wrote:
Thu Oct 17, 2019 7:28 pm
Gabriel A. Lozada wrote:
Thu Oct 17, 2019 2:36 pm
Such technical details aside, we've learned from Trurl Klapaucius is that TIAA's General Account quality is somewhat worse now than it was at year-end 2015.
I wonder whether this is due to the low interest rate environment. Since TIAA has contractual obligations to credit TIAA Traditional participants with at least 3% crediting rate, it has to find the yield somewhere.
TIAA has not publicly discussed changes in General Account quality, but I've found some information about general insurance industry trends.

One source a Financial Times article, "Search for yield draws US life insurers to risky places (Feb 21, 2019)" https://www.ft.com/content/7ddcd942-35c ... 2a211d90d5. This article discusses "Fitch Ratings: US Life Insurers' Investments More Risky Since Financial Crisis" (21 Feb 2019) https://www.fitchratings.com/site/pr/10062907. According to these sources, in general, insurance companies' portfolios have lower-rated debt now than before the Great Recession began. Part of the reason is because there is less high-quality corporate debt outstanding than there used to be. One of the people interviewed by the Financial Times believes that insurance industry portfolios are more diversified than they used to be, with more (taxable) municipal bonds, bank loans, private placements, and "structured products," so that "credit quality is down but default risk in portfolios is not higher."

Another source is "Positioning Life Insurance Portfolios for the Next Recession (and Beyond)" https://www.guggenheimpartners.com/pers ... -recession. Two of the charts there are particularly interesting. The first is entitled "Small Increases in Below Investment-Grade Holdings Can Put Capital at Risk"; it shows that below-investment-grade holdings were actually less on 12.31.2018 than they were in 2015. The second chart is entitled "NAIC 2 Holdings Have Surpassed Crisis Highs." The NAIC 2 rating is equivalent to BBB, and the chart shows that insurance industry holdings of it have risen since 2015.
Thanks for the summary.

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Fri Oct 18, 2019 1:55 pm

I just noticed that the URL given above for my paper is was pointing to an older version of it; the new version, with a slightly different title and some other changes, is at http://content.csbs.utah.edu/~lozada/Re ... TIAA_5.pdf. I'm sorry I didn't notice this earlier.

Edit: One now gets the new file even if one clicks on the old link.

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Fri Oct 18, 2019 5:08 pm

Trurl Klapaucius wrote:
Fri Oct 18, 2019 6:07 am
Given the “inconsistent” fashion TIAA Traditional crediting rates are reported, as described in the appendix of your paper, can the reported credit quality and duration of their holdings be trusted? In your study, were you able to perform an analysis similar to the one presented in the appendix to verify the consistency of reported credit quality and duration? Did this confirm that they were reported correctly?
The short answer is no, I didn't verify the consistency of TIAA's reported credit quality and duration.

The long answer is that the data on portfolio quality which I used came from the TIAA Annual Statement for the year ending 12/31/2015, which is this document https://www.tiaa.org/public/pdf/tiaa_an ... t_2015.pdf available on this web page: https://www.tiaa.org/public/about-tiaa/ ... nt-library (TIAA has more such material on its website now than it had some time ago.) In 2015, TIAA did not publish any documents for the general public which claimed that its General Account had any particular quality, so there was no consistency to check. One could do such a check for consistency for more recent years, but I haven't.

The data on portfolio duration which I used came from a document meant for the general public (TIAA (2016a) in the paper's bibliography; only the current version is now available at its URL). There is no easy way to check this for consistency with the TIAA Annual Statement because the annual statement only classifies bonds by general maturity ("over 1 year through 5 years," "over 5 years through 10 years," etc.), not by exact duration. Furthermore, TIAA's reported duration figure is (as it should be) option-adjusted, and to confirm it one would have to know the call features of TIAA's bonds. (Footnote: The Annual Statement does give exact maturity dates for "Replication (Synthetic Asset) Transactions Open as of December 31 of Current Year," which are credit default swaps, not bonds; see the 2015 Annual Statement's pages SI112 and SI112.1.)

The upshot is that I basically took TIAA's data at face value, and only discovered the discrepancy by accident. Ultimately, insurance companies don't have to tell their policyholders anything about their portfolio, and they have to tell insurance regulators a lot about their portfolio. I'm glad TIAA has begun to tell the public more about its portfolio, and one can hope that they will put enough effort into that direction that going forward there will not be further mistakes.

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Fri Oct 18, 2019 5:41 pm

dknightd wrote:
Fri Oct 18, 2019 7:49 am
The paper addresses the “public fixed income” portion of TIAA traditional. A significant part of Traditional's investments are not available to the public.
Now that I've actually remembered to put the current version of my paper on the web, instead of the outdated version, you are able to see in the current version's footnote 9 that the quality breakdown was for all of the General Account's bonds, both those publicly traded and those privately placed. However, it is true that as footnote 15 points out, only 85.914% of the General Account was invested in bonds; most of the rest, at 11.076%, was classified as "other," with the remaining 3.01% being equity, real estate, contract loans, and derivatives.

I've revised my Oct. 16 11:18pm response to Trurl Klapaucius for a second time to state this correctly (see the passages preceded by "EDIT").

Northern Flicker
Posts: 4917
Joined: Fri Apr 10, 2015 12:29 am

Re: Composition and safety of TIAA traditional annuity

Post by Northern Flicker » Fri Oct 18, 2019 8:05 pm

The question to answer is: are you more comfortable with liquidity risk or term risk? If bond volatility gives you jitters but you are comfortable with the liquidity restrictions of TIAA Trad then it is a good choice. If you don’t like liquidity restrictions and can focus on overall portfolio volatility instead of the volatility of a bond fund held, then a good quality low cost bond fund is probably a better choice.
Last edited by Northern Flicker on Fri Oct 18, 2019 10:34 pm, edited 1 time in total.
Index fund investor since 1987.

lassevirensghost
Posts: 96
Joined: Mon May 07, 2018 5:33 am

Re: Composition and safety of TIAA traditional annuity

Post by lassevirensghost » Fri Oct 18, 2019 8:46 pm

So should I stop putting fixes income contributions in Trad? Almost the entirety of that portion of my (smallish) portfolio is in Trad.
“Groucho, how do you invest your money?” | “All in bonds.” | “But Groucho, they don’t pay much return.” | “They do when you have a lot of em!”

student
Posts: 4133
Joined: Fri Apr 03, 2015 6:58 am

Re: Composition and safety of TIAA traditional annuity

Post by student » Fri Oct 18, 2019 8:51 pm

lassevirensghost wrote:
Fri Oct 18, 2019 8:46 pm
So should I stop putting fixes income contributions in Trad? Almost the entirety of that portion of my (smallish) portfolio is in Trad.
I don't think the paper suggests that you should stop putting money in there, it is just there is no free lunch, and it should not be one. I think Northern Flicker has a good summary.
Northern Flicker wrote:
Fri Oct 18, 2019 8:05 pm
The question to answer is: are you more comfortable with liquidity risk or term risk? If bond volatility gives you jitters but you are comfortable with the liquidity restrictions of TIAA Trad then it is a good choice. If you don’t like liquidity restrictions and can focus on overall portfolio volatility instead of the volatility of a bond fund held, then a good quality low cost bond fund is probably a better choice:
Full disclosure: I use TIAA Traditional almost exclusively for my non-cash portion of fixed income.
Last edited by student on Fri Oct 18, 2019 9:11 pm, edited 1 time in total.

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Fri Oct 18, 2019 9:04 pm

dknightd wrote:
Fri Oct 18, 2019 8:53 am
Gabriel A. Lozada wrote:
Thu Oct 17, 2019 8:02 pm
Traditional was second-degree stochastic-dominated by several of the alternatives: the risk reduction offered by Traditional would not have been worth it to any risk-averse expected-utility-maximizing investor.
Can you explain what a "risk-averse expected-utility-maximizing investor" is?
Thanks


Edit: Also I'm curious, you work at a University so likely qualify to buy TIAA Traditional contracts, do you own any? Why, or why not?
Again, Thanks
Wikipedia has an explanation of the expected utility hypothesis at https://en.wikipedia.org/wiki/Expected_ ... hypothesis, but here's a simple example. Suppose a person's "utility" (happiness) depends on their income "m" according to the formula "square root of m," that is, sqrt(m). For example, if the person gets $5 their utility is sqrt($5) = 2.23607, and if the person instead got $15 their utility is sqrt($15) = 3.87298. Suppose this person is faced with a 50% probability of receiving $5 and a 50% probability of receiving $15. If the person's utility level when faced with this uncertain outcome is equal to (1/2) sqrt($5) + (1/2) sqrt($15) = 3.05453, then the person's preferences obey the expected utility hypothesis. If the person's utility level when faced with this uncertain outcome is not equal to 3.05453 then the person's preferences would not obey the expected utility hypothesis. A person's preferences will satisfy the expected utility hypothesis if and only if those preferences obey the "four axioms of von Neumann-Morgenstern rationality" (see https://en.wikipedia.org/wiki/Von_Neuma ... ty_theorem). Many economists find it quite plausible that people's choices under uncertainty do obey those axioms, and so much of economics and finance is based on assuming that people are expected-utility maximizers. However, psychologists Daniel Kahneman and Amos Tversky famously showed that many people sometimes do not behave that way, giving rise to "behavioral economics" and "behavioral finance"; for more on this, see the "Criticism" section of the first Wikipedia article I cited.

Risk aversion also has a Wikipedia article, https://en.wikipedia.org/wiki/Risk_aversion, so I'll just illustrate with the example of the previous paragraph, the example of the person facing 50% odds of receiving $5 and 50% odds of receiving $15. If this person would prefer to receive (1/2)*$5 + (1/2)*15 = $10 for sure, instead of receiving the uncertain outcome, then we say that this person is "risk averse."

Assuming that investors are risk-averse expected-utility-maximizers is probably the most common assumption which economists make concerning investor preferences, although the Kahneman/Tversky, "behavioral economics" critique is considered important by a not-insignificant group of other economists.

As to me personally, yes, I am eligible to buy Traditional contracts. I don't, even though the General Account's quality and duration are probably quite reasonable and prudent, because, as outlined in the paper, the past risk-adjusted performance hasn't been better than quality- and duration-matched alternatives. (Also because liability matching is attractive, and that would require, at my age, durations longer than 8 years, but this would not be a good place to get sidetracked into a discussion of liability matching.) However I might invest in Traditional in the future, say, to get annuity income after retirement. That would require comparing a Traditional-based single-premium annuity to alternative SPIAs, which is quite a different comparison than the one done in the paper. And I should say that although there have been a few alternatives to Traditional for accumulating fixed-income monies which have done better than Traditional, there are very, very many more fixed-income alternatives which have done worse than Traditional.

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Sat Oct 19, 2019 12:07 am

dknightd wrote:
Fri Oct 18, 2019 9:34 am
TIAA Traditional was designed to provide educators with a stable and secure retirement. Without those educators spending time to get educated about investing. It was only later that TIAA offered other investment options.
Having read your entire paper now, my take is TIAA Traditional provides "no free lunch", but it also helps prevent a bully from taking your lunch money.
This is insightful.

There are social costs in setting up retirement systems so complicated that it behooves ordinary workers to "spend time to get educated about investing," but a full analysis would have to include whatever benefits such systems might have as well. I wonder whether educators who invested with TIAA in the days when TIAA only offered two choices, CREF Stock and Traditional, have done better or worse than younger generations of educators who have had many more choices. However it would probably be difficult to get the data to investigate this, and thought would have to go into just how to pose the question since the passage of time has seen other changes in addition to the proliferation of investment choices.

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Sat Oct 19, 2019 1:25 am

crefwatch wrote:
Fri Oct 18, 2019 12:05 pm
If you are afraid that your TIAA Traditional money will disappear, then you do care about prudent investing by this New York State regulated insurance company. But millions of people buy retirement investments from less highly-rated insurance companies every day. TIAA Traditional is a contract with the investor for TIAA to perform certain financial acts for them. It is not a Separate Account, and it is not fractional ownership of anything. It certainly depends on the General Account's success to not disappear, but it does not depend on the General Account's success to pay next year's promised benefits.
This is correct, as far as it goes. However, given Traditional's withdrawal restrictions, its investors will be concerned with many years beyond next year.

Page 5 of my paper says: "While it is true that, as TIAA (2018a p. 6) says, '[Traditional’s] participants do not invest in the TIAA general account portfolio' and the rate of interest 'is determined at the discretion of TIAA’s Board of Trustees' (op. cit. p. 5) rather than directly from the underlying holdings, TIAA’s General Account’s investments are what make Traditional possible and so are what someone trying to either benchmark or mimic Traditional using other instruments would want to imitate." I stand by this statement.
crefwatch wrote:
Fri Oct 18, 2019 12:05 pm
To the average lay investor, it is misleading to compare TIAA Traditional to Open-End Bond Mutual Funds. It is also misleading to think about Open-End Mutual Funds and the large amount of TIAA Traditional that is locked up for nine years and a day as if they were comparable obligations of a financial-services provider.
As I report on p. 4, a comparison TIAA has made is: "A 10-year risk-free rate seems appropriate to use as a benchmark for the TIAA Traditional Annuity." That is hardly an appropriate comparison, for the reasons given in the paper. In another place, TIAA has written, "TIAA Traditional isn't structured exactly the same way as guaranteed annuities that other financial services firms may offer, so it's difficult to make a direct apples-to-apples comparison," and then gives no benchmark at all. Nevertheless, Traditional's performance ultimately derives from the performance of TIAA's General Account, and if that theoretical assertion isn't convincing, the empirical results shown in Figure 3 should be: in the long-run, Traditional has behaved just like a somewhat smoothed, often somewhat-lower-returning bond fund of the appropriate quality and duration.

It is certainly misleading to think about open-end mutual funds and the variety of TIAA Traditional that is locked up for nine years and a day as if they were identical obligations of a financial-services provider. That is why I remind the reader that “[Traditional’s] participants do not invest in the TIAA general account portfolio” and the rate of interest “is determined at the discretion of TIAA’s Board of Trustees.” However I think the paper demonstrates that these investments, if liquidated gradually over nine years, are comparable.
crefwatch wrote:
Fri Oct 18, 2019 12:05 pm
Clearly, most of the posters here know the difference, but for less sophisticated, or more casual readers of this thread, I must point out that the title of the scholarly paper is way off of "accurate." It is completely wrong to equate a consumer (investor's) view of the TIAA Traditional Annuity product with the structure of the TIAA General Account....Sure, it's sensible to ask whether TIAA might go belly-up after 100 years, especially since the financial world (and the company) have changed so much since 1987. But I repeat that the title of the paper is a disservice to the scholarship.
The title of the revised version of the paper, which you now get when clicking on any of the above links pointing to the paper, is "Fixed Income for Retirement Saving: TIAA Traditional’s Lessons on Quality, Duration, Risk, and Gradual Withdrawals."

I'm not sure what about the title could be considered inaccurate. Of course it is wrong to equate an investor's view of Traditional with the structure of the General Account; if those two were the same, investors would already understand the General Account and there would have been no point in writing the paper. The paper is an attempt to go beyond a consumer's view of TIAA Traditional in order to come up with an economist's view of Traditional, and the paper is interesting to the extent that the economist's view is novel and different from the consumer's view. The consumer's view, or at least the view of TIAA's marketing materials, is that TIAA is in a unique class of high-return, low-risk investments. This economist's view is that, in the long run, that view is wrong: what TIAA gave to a consumer is quite comparable to what a particular combination of bond funds would have given to the consumer, but slightly worse.

The paper never discusses "whether TIAA might go belly-up'' except in one footnote (19), where the point is exactly Mr./Ms. CrefWatch's point: Traditional is not a "comparable obligation'' to a mutual fund.

Overall, I'm at a loss to understand why "the title of the paper is a disservice to the scholarship," but perhaps you'll want to explain it some more.

student
Posts: 4133
Joined: Fri Apr 03, 2015 6:58 am

Re: Composition and safety of TIAA traditional annuity

Post by student » Sat Oct 19, 2019 5:56 am

Gabriel A. Lozada wrote:
Sat Oct 19, 2019 12:07 am
I wonder whether educators who invested with TIAA in the days when TIAA only offered two choices, CREF Stock and Traditional, have done better or worse than younger generations of educators who have had many more choices. However it would probably be difficult to get the data to investigate this, and thought would have to go into just how to pose the question since the passage of time has seen other changes in addition to the proliferation of investment choices.
I guess in the old days, there were few funds with decent expense ratios, and this made TIAA even more attractive. Given that Bernake invested in TIAA Traditional (when it was disclosed in 2007), it is probably safe to say that while TIAA Traditional may not be the best choice for fixed income, it is a good enough choice for many. Thank you again for answering our questions.

dknightd
Posts: 1860
Joined: Wed Mar 07, 2018 11:57 am

Re: Composition and safety of TIAA traditional annuity

Post by dknightd » Sat Oct 19, 2019 7:19 am

Gabriel A. Lozada wrote:
Fri Oct 18, 2019 9:04 pm

Wikipedia has an explanation of the expected utility hypothesis at https://en.wikipedia.org/wiki/Expected_ ... hypothesis, but here's a simple example. Suppose a person's "utility" (happiness) depends on their income "m" according to the formula "square root of m," that is, sqrt(m). For example, if the person gets $5 their utility is sqrt($5) = 2.23607, and if the person instead got $15 their utility is sqrt($15) = 3.87298. Suppose this person is faced with a 50% probability of receiving $5 and a 50% probability of receiving $15. If the person's utility level when faced with this uncertain outcome is equal to (1/2) sqrt($5) + (1/2) sqrt($15) = 3.05453, then the person's preferences obey the expected utility hypothesis. If the person's utility level when faced with this uncertain outcome is not equal to 3.05453 then the person's preferences would not obey the expected utility hypothesis. A person's preferences will satisfy the expected utility hypothesis if and only if those preferences obey the "four axioms of von Neumann-Morgenstern rationality" (see https://en.wikipedia.org/wiki/Von_Neuma ... ty_theorem).
Thanks for the explanation, and links. I'm still not sure I completely understand it, but it is starting to make sense, and I have more to study now. Obviously I'm not an economist . . .
Gabriel A. Lozada wrote:
Fri Oct 18, 2019 9:04 pm
Risk aversion also has a Wikipedia article, https://en.wikipedia.org/wiki/Risk_aversion, so I'll just illustrate with the example of the previous paragraph, the example of the person facing 50% odds of receiving $5 and 50% odds of receiving $15. If this person would prefer to receive (1/2)*$5 + (1/2)*15 = $10 for sure, instead of receiving the uncertain outcome, then we say that this person is "risk averse."
I'd probably prefer the guaranteed $10. That would give me a utility of sqrt($10)=3.16, which is larger than (1/2) sqrt($5) + (1/2) sqrt($15) = 3.05453

So, I guess that makes me a risk-averse expected-utility-maximizer investor.

Thanks for taking the time to help educate us :)

edit: now I guess I have to define how I see the "utility" of money ;)

Trurl Klapaucius
Posts: 53
Joined: Tue Aug 06, 2013 1:11 pm

Re: Composition and safety of TIAA traditional annuity

Post by Trurl Klapaucius » Sat Oct 19, 2019 8:06 am

dknightd wrote:
Fri Oct 18, 2019 7:49 am
The paper addresses the “public fixed income” portion of TIAA traditional.
In an attempt to curb propagation of my misrepresentation of Lozada’s paper, I’ve edited my initial two posts in this thread to change “public fixed income” to “bond”. Sorry for the confusion I caused.

crefwatch
Posts: 263
Joined: Sun Apr 15, 2007 1:07 pm
Location: New Jersey, USA

Re: Composition and safety of TIAA traditional annuity

Post by crefwatch » Sat Oct 19, 2019 8:16 am

I thank Mr. Lozada for his substantive (rather than fragmentary, as often happens on newsboards) quotes of my comments, which represent my posting very fairly. What I meant by disliking the old title of the paper was that valuable scholarship has been done, but the vast number of casual internet readers can get misled by two-sentence summaries they read around newsboards. Most internet readers do not read as intensely as the majority of posters in this thread. (Ironica Typeface: And most investors can barely construct a coherent asset allocation, let alone build a replica of TIAA Traditional.)

In 1975, when I became eligible for TIAA-CREF in my second or third year of employment, it was the ONLY retirement plan available at my smaller, prestigious, educational institution. Since it had 5% employer-matching, it was a no-brainer. CREF [Stock] was a U.S. stock index fund with an ER of 0.22%, and I was prohibited from allocating more than half of my monthly Salary Reduction to CREF! Disclosures made it clear that I would NEVER be able to receive any of my contributions back in any way other than a lifetime annuity. I signed up anyway. (Of course, that restriction eventually disappeared, and a good thing.)

bs1
Posts: 23
Joined: Sun Jun 24, 2012 7:10 pm

Re: Composition and safety of TIAA traditional annuity

Post by bs1 » Sat Oct 19, 2019 8:46 am

Gabriel,

long-term nominal bonds suffer a lot when the inflation rate rises unexpectedly, owing to capital losses and reduced purchasing power. the period studied in the paper looks like one of declining interest rates and low inflation. would Traditional comparatively fare better during an interest rate spike during the (10-year) withdrawal phase? in other words, does Traditional’s price stability mitigate interest rate risk, potentially increasing its appeal?

thanks for you time.

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Sat Oct 19, 2019 6:02 pm

vineviz wrote:
Fri Oct 18, 2019 12:12 pm
The point of the research, which is both relevant and accurate, is to explore the conditions under which normal investors might be able to replicate the returns of TIAA “annuity”.

The finding that such replication isn’t especially difficult, at least economically, should IMHO be of interest to investors who might otherwise contort their investment plans to include TIAA (or other stable value funds, like the TSP’s G Fund).
Yes, thanks!

User avatar
dodecahedron
Posts: 4808
Joined: Tue Nov 12, 2013 12:28 pm

Re: Composition and safety of TIAA traditional annuity

Post by dodecahedron » Sat Oct 19, 2019 6:25 pm

student wrote:
Sat Oct 19, 2019 5:56 am
Gabriel A. Lozada wrote:
Sat Oct 19, 2019 12:07 am
I wonder whether educators who invested with TIAA in the days when TIAA only offered two choices, CREF Stock and Traditional, have done better or worse than younger generations of educators who have had many more choices. However it would probably be difficult to get the data to investigate this, and thought would have to go into just how to pose the question since the passage of time has seen other changes in addition to the proliferation of investment choices.
I guess in the old days, there were few funds with decent expense ratios, and this made TIAA even more attractive. Given that Bernake invested in TIAA Traditional (when it was disclosed in 2007), it is probably safe to say that while TIAA Traditional may not be the best choice for fixed income, it is a good enough choice for many. Thank you again for answering our questions.
I have talked to a number of old timers (from the days when there was just a simple stark choice, one flavor of TIAA Trad and one flavor of CREF, with a limit of 50% on the latter) who have seemed quite content with how things worked out. They do note that they did not have to spend a lot of time evaluating the many options that younger folks subsequently had. I will note that quite a few colleges subsequently used to have overwhelming menus with hundreds (yes, hundreds!) of funds that one could choose. Many people from *that* era found themselves so overwhelmed by the choices that they felt the need to hire expensive advisers to help them choose from their menu options. (And the advisers often recommended a confusing mish-mash of choices with little rhyme nor reason to them.) Other folks just pretty much seemed to throw darts at their choices.

columbia
Posts: 1978
Joined: Tue Aug 27, 2013 5:30 am

Re: Composition and safety of TIAA traditional annuity

Post by columbia » Sat Oct 19, 2019 6:34 pm

crefwatch wrote:
Fri Oct 18, 2019 12:05 pm
Trurl Klapaucius wrote:
Wed Oct 16, 2019 5:09 pm
Bonds for Retirement Saving: TIAA Traditional’s Lessons on Quality, Duration, Risk, and Gradual Withdrawals by Gabriel A. Lozada Department of Economics, University of Utah Salt Lake City, UT 84112, USA http://content.csbs.utah.edu/~lozada/Re ... A_4_UU.pdf
Clearly, most of the posters here know the difference, but for less sophisticated, or more casual readers of this thread, I must point out that the title of the scholarly paper is way off of "accurate." It is completely wrong to equate a consumer (investor's) view of the TIAA Traditional Annuity product with the structure of the TIAA General Account.

If you are afraid that your TIAA Traditional money will disappear, then you do care about prudent investing by this New York State regulated insurance company. But millions of people buy retirement investments from less highly-rated insurance companies every day. TIAA Traditional is a contract with the investor for TIAA to perform certain financial acts for them. It is not a Separate Account, and it is not fractional ownership of anything. It certainly depends on the General Account's success to not disappear, but it does not depend on the General Account's success to pay next year's promised benefits.

To the average lay investor, it is misleading to compare TIAA Traditional to Open-End Bond Mutual Funds. It is also misleading to think about Open-End Mutual Funds and the large amount of TIAA Traditional that is locked up for nine years and a day as if they were comparable obligations of a financial-services provider.

Between 1928 and 1941 [Greenough ... ] TIAA made multiple changes to the annuity structure to reflect better knowledge of actuarial realities. But there was a crisis in 1948 over obligations incurred (to us, that is) before 1936. Between a large grant from the Carnegie Corporation, and increasing reserves at TIAA, those investors were protected. It's wrong to assume that exact protection could happen again, especially since the company is orders-of-magnitude larger than it was in 1948. But I don't lie awake at night worrying about the General Account!

Sure, it's sensible to ask whether TIAA might go belly-up after 100 years, especially since the financial world (and the company) have changed so much since 1987. But I repeat that the title of the paper is a disservice to the scholarship.

The appropriate comparisons would be other stable value funds (presuming one has access to TIAA + at least one other).

student
Posts: 4133
Joined: Fri Apr 03, 2015 6:58 am

Re: Composition and safety of TIAA traditional annuity

Post by student » Sat Oct 19, 2019 6:35 pm

dodecahedron wrote:
Sat Oct 19, 2019 6:25 pm
student wrote:
Sat Oct 19, 2019 5:56 am
Gabriel A. Lozada wrote:
Sat Oct 19, 2019 12:07 am
I wonder whether educators who invested with TIAA in the days when TIAA only offered two choices, CREF Stock and Traditional, have done better or worse than younger generations of educators who have had many more choices. However it would probably be difficult to get the data to investigate this, and thought would have to go into just how to pose the question since the passage of time has seen other changes in addition to the proliferation of investment choices.
I guess in the old days, there were few funds with decent expense ratios, and this made TIAA even more attractive. Given that Bernake invested in TIAA Traditional (when it was disclosed in 2007), it is probably safe to say that while TIAA Traditional may not be the best choice for fixed income, it is a good enough choice for many. Thank you again for answering our questions.
I have talked to a number of old timers (from the days when there was just a simple stark choice, one flavor of TIAA Trad and one flavor of CREF, with a limit of 50% on the latter) who have seemed quite content with how things worked out. They do note that they did not have to spend a lot of time evaluating the many options that younger folks subsequently had. I will note that quite a few colleges subsequently used to have overwhelming menus with hundreds (yes, hundreds!) of funds that one could choose. Many people from *that* era found themselves so overwhelmed by the choices that they felt the need to hire expensive advisers to help them choose from their menu options. (And the advisers often recommended a confusing mish-mash of choices with little rhyme nor reason to them.) Other folks just pretty much seemed to throw darts at their choices.
My university offers TIAA and Fidelity. For Fidelity, we have "everything," as far as I can tell, almost every available Fidelity fund. Like your colleagues, my older friends who only have TIAA and CREF are in general happy with them.

student
Posts: 4133
Joined: Fri Apr 03, 2015 6:58 am

Re: Composition and safety of TIAA traditional annuity

Post by student » Sat Oct 19, 2019 6:36 pm

columbia wrote:
Sat Oct 19, 2019 6:34 pm
crefwatch wrote:
Fri Oct 18, 2019 12:05 pm
Trurl Klapaucius wrote:
Wed Oct 16, 2019 5:09 pm
Bonds for Retirement Saving: TIAA Traditional’s Lessons on Quality, Duration, Risk, and Gradual Withdrawals by Gabriel A. Lozada Department of Economics, University of Utah Salt Lake City, UT 84112, USA http://content.csbs.utah.edu/~lozada/Re ... A_4_UU.pdf
Clearly, most of the posters here know the difference, but for less sophisticated, or more casual readers of this thread, I must point out that the title of the scholarly paper is way off of "accurate." It is completely wrong to equate a consumer (investor's) view of the TIAA Traditional Annuity product with the structure of the TIAA General Account.

If you are afraid that your TIAA Traditional money will disappear, then you do care about prudent investing by this New York State regulated insurance company. But millions of people buy retirement investments from less highly-rated insurance companies every day. TIAA Traditional is a contract with the investor for TIAA to perform certain financial acts for them. It is not a Separate Account, and it is not fractional ownership of anything. It certainly depends on the General Account's success to not disappear, but it does not depend on the General Account's success to pay next year's promised benefits.

To the average lay investor, it is misleading to compare TIAA Traditional to Open-End Bond Mutual Funds. It is also misleading to think about Open-End Mutual Funds and the large amount of TIAA Traditional that is locked up for nine years and a day as if they were comparable obligations of a financial-services provider.

Between 1928 and 1941 [Greenough ... ] TIAA made multiple changes to the annuity structure to reflect better knowledge of actuarial realities. But there was a crisis in 1948 over obligations incurred (to us, that is) before 1936. Between a large grant from the Carnegie Corporation, and increasing reserves at TIAA, those investors were protected. It's wrong to assume that exact protection could happen again, especially since the company is orders-of-magnitude larger than it was in 1948. But I don't lie awake at night worrying about the General Account!

Sure, it's sensible to ask whether TIAA might go belly-up after 100 years, especially since the financial world (and the company) have changed so much since 1987. But I repeat that the title of the paper is a disservice to the scholarship.

The appropriate comparisons would be other stable value funds (presuming one has access to TIAA + at least one other).
At my university, TIAA Traditional is the only stable value fund available.

columbia
Posts: 1978
Joined: Tue Aug 27, 2013 5:30 am

Re: Composition and safety of TIAA traditional annuity

Post by columbia » Sat Oct 19, 2019 6:39 pm

student wrote:
Sat Oct 19, 2019 6:36 pm
columbia wrote:
Sat Oct 19, 2019 6:34 pm
crefwatch wrote:
Fri Oct 18, 2019 12:05 pm
Trurl Klapaucius wrote:
Wed Oct 16, 2019 5:09 pm
Bonds for Retirement Saving: TIAA Traditional’s Lessons on Quality, Duration, Risk, and Gradual Withdrawals by Gabriel A. Lozada Department of Economics, University of Utah Salt Lake City, UT 84112, USA http://content.csbs.utah.edu/~lozada/Re ... A_4_UU.pdf
Clearly, most of the posters here know the difference, but for less sophisticated, or more casual readers of this thread, I must point out that the title of the scholarly paper is way off of "accurate." It is completely wrong to equate a consumer (investor's) view of the TIAA Traditional Annuity product with the structure of the TIAA General Account.

If you are afraid that your TIAA Traditional money will disappear, then you do care about prudent investing by this New York State regulated insurance company. But millions of people buy retirement investments from less highly-rated insurance companies every day. TIAA Traditional is a contract with the investor for TIAA to perform certain financial acts for them. It is not a Separate Account, and it is not fractional ownership of anything. It certainly depends on the General Account's success to not disappear, but it does not depend on the General Account's success to pay next year's promised benefits.

To the average lay investor, it is misleading to compare TIAA Traditional to Open-End Bond Mutual Funds. It is also misleading to think about Open-End Mutual Funds and the large amount of TIAA Traditional that is locked up for nine years and a day as if they were comparable obligations of a financial-services provider.

Between 1928 and 1941 [Greenough ... ] TIAA made multiple changes to the annuity structure to reflect better knowledge of actuarial realities. But there was a crisis in 1948 over obligations incurred (to us, that is) before 1936. Between a large grant from the Carnegie Corporation, and increasing reserves at TIAA, those investors were protected. It's wrong to assume that exact protection could happen again, especially since the company is orders-of-magnitude larger than it was in 1948. But I don't lie awake at night worrying about the General Account!

Sure, it's sensible to ask whether TIAA might go belly-up after 100 years, especially since the financial world (and the company) have changed so much since 1987. But I repeat that the title of the paper is a disservice to the scholarship.

The appropriate comparisons would be other stable value funds (presuming one has access to TIAA + at least one other).
At my university, TIAA Traditional is the only stable value fund available.
Yes. I doubt that many have access to more than one.

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Sat Oct 19, 2019 6:54 pm

student wrote:
Fri Oct 18, 2019 8:51 pm
lassevirensghost wrote:
Fri Oct 18, 2019 8:46 pm
So should I stop putting fixes income contributions in Trad? Almost the entirety of that portion of my (smallish) portfolio is in Trad.
I don't think the paper suggests that you should stop putting money in there, it is just there is no free lunch, and it should not be one. I think Northern Flicker has a good summary.
Northern Flicker wrote:
Fri Oct 18, 2019 8:05 pm
The question to answer is: are you more comfortable with liquidity risk or term risk? If bond volatility gives you jitters but you are comfortable with the liquidity restrictions of TIAA Trad then it is a good choice. If you don’t like liquidity restrictions and can focus on overall portfolio volatility instead of the volatility of a bond fund held, then a good quality low cost bond fund is probably a better choice:
I agree with student: the paper does not suggest not investing in Traditional, just that there was no free lunch, and it was possible to get somewhat better results in the long run (definitely not in the short run!) using duration- and quality-matched bond funds. Again, the short-run results of Traditional will be very different from the short-run results of the duration- and quality-matched bond funds, and by "different" I mean "better" (usually) because bond funds will have many more short-run fluctuations, including many times when they lose money, as the paper's Figure 2 shows. To echo Northern Flicker, if the money-losing periods that will come when investing in roughly-8-year-duration bond funds are going to cause anxiety, Traditional is the better choice! I agree with Northern Flicker's last sentence too, but I'd like to add another sentence which is more in the spirit of my paper: if you don’t like liquidity restrictions and can focus on long-run (say, 10-year) bond fund volatility instead short-run (monthly or annual) bond fund volatility, then a good quality low cost bond fund of similar quality and duration, if available, is probably a better choice.

To clarify, by "long-run (say, 10-year) volatility" I don't mean "monthly or weekly or annual volatility over 10 years." A simplified example of what I mean is: write down what the rate of return was from 2000 to 2010; write down the rate of return from 2001 to 2011; write down the rate of return from 2002 to 2012; then find the variance of the three numbers you wrote down. To get a meaningful number for volatility you hope to have more than three numbers, but this illustrates the kind of long-run thinking about bond volatility which it'd be useful for long-term investors to develop. The paper does something more complicated, looking the ex-post feasible withdrawal amount over successive 9 year periods, but the flavor is the same.

(As I wrote above, "if one's fixed-income investments are undertaken for reasons alien to those considered in the paper, such as to attempt to counteract short-term movements in the equity markets, then very different considerations would enter and even the daily fluctuations of a bond fund could have importance." In other words, there are some long-term investors for whom short-run bond volatility is and should be important. Different strokes for different folks.)

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Sat Oct 19, 2019 7:02 pm

student wrote:
Sat Oct 19, 2019 5:56 am
I guess in the old days, there were few funds with decent expense ratios, and this made TIAA even more attractive. Given that Bernake invested in TIAA Traditional (when it was disclosed in 2007), it is probably safe to say that while TIAA Traditional may not be the best choice for fixed income, it is a good enough choice for many.
Yes, absolutely.
student wrote:
Sat Oct 19, 2019 5:56 am
Thank you again for answering our questions.
You're very welcome; thank you for your comments!

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Sat Oct 19, 2019 7:53 pm

dknightd wrote:
Sat Oct 19, 2019 7:19 am
Gabriel A. Lozada wrote:
Fri Oct 18, 2019 9:04 pm

Wikipedia has an explanation of the expected utility hypothesis at https://en.wikipedia.org/wiki/Expected_ ... hypothesis [...]
Thanks for the explanation, and links. I'm still not sure I completely understand it, but it is starting to make sense, and I have more to study now. Obviously I'm not an economist . . .
Gabriel A. Lozada wrote:
Fri Oct 18, 2019 9:04 pm
Risk aversion also has a Wikipedia article, https://en.wikipedia.org/wiki/Risk_aversion, so I'll just illustrate with the example of the previous paragraph, the example of the person facing 50% odds of receiving $5 and 50% odds of receiving $15. If this person would prefer to receive (1/2)*$5 + (1/2)*15 = $10 for sure, instead of receiving the uncertain outcome, then we say that this person is "risk averse."
I'd probably prefer the guaranteed $10. That would give me a utility of sqrt($10)=3.16, which is larger than (1/2) sqrt($5) + (1/2) sqrt($15) = 3.05453

So, I guess that makes me a risk-averse expected-utility-maximizer investor.
It certainly makes you risk-averse, but deciding whether you judge uncertain situations using expected utility is trickier.

For example, from https://en.wikipedia.org/wiki/Von_Neuma ... ty_theorem, one of the von Neumann-Morgenstern axioms (the completeness axiom) is: either an investor prefers investment "L" to investment "M," or the investor prefers investment "M" to investment "L," or the investor is indifferent between investments "L" and "M." This essentially means investors never change their minds: they never switch from preferring "L" over "M" to preferring "M" over "L." While this assumption makes it easier for economists to study behavior under uncertainty just using math, pencil and paper, without needing to study any actual humans, if one does look at actual human behavior it doesn't always obey this axiom.

To apply this to the paper, over the historical period studied, with perfect foresight of the probability distributions (but not the actual outcomes), a risk-averse expected-utility-maximizing investor would have preferred some of the alternatives to Traditional, but other kinds of investors might well have preferred Traditional. You might want to look at Figure 3 and Figure 5 and decide for yourself which investment you would have preferred. If you decide you would have preferred Traditional, that's OK, it just means your preferences don't obey all of the von Neumann-Morgenstern axioms (or that you're risk-loving).
dknightd wrote:
Sat Oct 19, 2019 7:19 am
Thanks for taking the time to help educate us :)
Thank you for your questions and comments!

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Sat Oct 19, 2019 8:08 pm

Trurl Klapaucius wrote:
Sat Oct 19, 2019 8:06 am
dknightd wrote:
Fri Oct 18, 2019 7:49 am
The paper addresses the “public fixed income” portion of TIAA traditional.
In an attempt to curb propagation of my misrepresentation of Lozada’s paper, I’ve edited my initial two posts in this thread to change “public fixed income” to “bond”. Sorry for the confusion I caused.
Thank you very much for your editing, as well as for your comments on the paper. If some weeks ago I had remembered to replace the paper's old version with the new version on my web site, even I would have been less confused. :oops:

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Sat Oct 19, 2019 11:01 pm

crefwatch wrote:
Sat Oct 19, 2019 8:16 am
I thank Mr. Lozada for his substantive (rather than fragmentary, as often happens on newsboards) quotes of my comments, which represent my posting very fairly. What I meant by disliking the old title of the paper was that valuable scholarship has been done, but the vast number of casual internet readers can get misled by two-sentence summaries they read around newsboards. Most internet readers do not read as intensely as the majority of posters in this thread. (Ironica Typeface: And most investors can barely construct a coherent asset allocation, let alone build a replica of TIAA Traditional.)

In 1975, when I became eligible for TIAA-CREF in my second or third year of employment, it was the ONLY retirement plan available at my smaller, prestigious, educational institution. Since it had 5% employer-matching, it was a no-brainer. CREF [Stock] was a U.S. stock index fund with an ER of 0.22%, and I was prohibited from allocating more than half of my monthly Salary Reduction to CREF! Disclosures made it clear that I would NEVER be able to receive any of my contributions back in any way other than a lifetime annuity. I signed up anyway. (Of course, that restriction eventually disappeared, and a good thing.)
Thank you for clarifying what you meant. Several weeks ago a journal editor suggested changing the title of the paper and I'm very glad I accepted his suggestion. Also, you might be interested to learn that until 2015, defined contribution retirement plans in the United Kingdom imposed forced annuitization, and in the Netherlands, some plans still force annuitization, at least as of the January 2015 date of this article: https://retirementincomejournal.com/art ... uum-in-uk/.

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Sat Oct 19, 2019 11:16 pm

bs1 wrote:
Sat Oct 19, 2019 8:46 am
Gabriel,

long-term nominal bonds suffer a lot when the inflation rate rises unexpectedly, owing to capital losses and reduced purchasing power. the period studied in the paper looks like one of declining interest rates and low inflation. would Traditional comparatively fare better during an interest rate spike during the (10-year) withdrawal phase? in other words, does Traditional’s price stability mitigate interest rate risk, potentially increasing its appeal?

thanks for you time.
Probably not in the long run. In the long run, Traditional cannot sustainably pay more than TIAA's General Account earns. During a period of rising interest rates, the General Account's value will fall and its yield will rise. This will be reflected in Traditional in no raising of its yield initially, and Traditional's eventual yield increases, when they come, will likely lag those of the General Account and of the broader market for a long time.

Now the short-run behavior of Traditional will be very much better than the short-run behavior of an intermediate- to long-term bond fund when interest rates go up, but short-run behavior is not what my paper's about.

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Sat Oct 19, 2019 11:40 pm

dodecahedron wrote:
Sat Oct 19, 2019 6:25 pm
student wrote:
Sat Oct 19, 2019 5:56 am
Gabriel A. Lozada wrote:
Sat Oct 19, 2019 12:07 am
I wonder whether educators who invested with TIAA in the days when TIAA only offered two choices, CREF Stock and Traditional, have done better or worse than younger generations of educators who have had many more choices. However it would probably be difficult to get the data to investigate this, and thought would have to go into just how to pose the question since the passage of time has seen other changes in addition to the proliferation of investment choices.
I guess in the old days, there were few funds with decent expense ratios, and this made TIAA even more attractive. Given that Bernake invested in TIAA Traditional (when it was disclosed in 2007), it is probably safe to say that while TIAA Traditional may not be the best choice for fixed income, it is a good enough choice for many. Thank you again for answering our questions.
I have talked to a number of old timers (from the days when there was just a simple stark choice, one flavor of TIAA Trad and one flavor of CREF, with a limit of 50% on the latter) who have seemed quite content with how things worked out. They do note that they did not have to spend a lot of time evaluating the many options that younger folks subsequently had. I will note that quite a few colleges subsequently used to have overwhelming menus with hundreds (yes, hundreds!) of funds that one could choose. Many people from *that* era found themselves so overwhelmed by the choices that they felt the need to hire expensive advisers to help them choose from their menu options. (And the advisers often recommended a confusing mish-mash of choices with little rhyme nor reason to them.) Other folks just pretty much seemed to throw darts at their choices.
student wrote:
Sat Oct 19, 2019 6:35 pm
Like your colleagues, my older friends who only have TIAA and CREF are in general happy with them.
Thank you for sharing these recollections. Conventional wisdom among most economists is that more choice is always better, but in the context of retirement plans, social psychologists have questioned this, and recent research by economists backs up the social psychologists' position. For example, see the 2016 working paper "Simplifying Choices In Defined Contribution Retirement Plan Design" by Keim and Mitchell, available from https://www.nber.org/papers/w21854.pdf, or its 2018 published version at https://www.cambridge.org/core/journals ... 3E7BB4308F (paywalled).

User avatar
dodecahedron
Posts: 4808
Joined: Tue Nov 12, 2013 12:28 pm

Re: Composition and safety of TIAA traditional annuity

Post by dodecahedron » Sun Oct 20, 2019 8:34 am

Gabriel A. Lozada wrote:
Sat Oct 19, 2019 11:16 pm
In the long run, Traditional cannot sustainably pay more than TIAA's General Account earns. During a period of rising interest rates, the General Account's value will fall and its yield will rise. This will be reflected in Traditional in no raising of its yield initially, and Traditional's eventual yield increases, when they come, will likely lag those of the General Account and of the broader market for a long time.
Your analysis does not reflect the fact that the General Account is not just set up to support the inflows and outflows to the very narrow and specific type of Traditional RA accounts that you analyzed.

The General Account stands behind many different types of TIAA liabilities (including for example, TIAA Life Insurance, both term and cash surrender value life and retained asset accounts, Supplemental Retirement Annuity (SRA) Trad, Retirement Choice (RC) Trad, IRA Trad, just to name a few.) The General Account is also the standby source of liquidity for TIAA Real Estate Annuity (TREA) redemptions. (TREA pays the GA a non-trivial sum for this liquidity guarantee, by the way.) And of course the General Account stands behind annuities in payout mode.

My point is that the operating margins and durations/term structures of all these many very different types of claims against the General Account are very different from the 9 year redemption provision of TIAA RA Trad. It is not appropriate to analyze the performance of the GA as if it only faces a 9 year redemption schedule.

Edited to add: my impression is that TIAA is also quite concerned about the sustainability of their products with 3% minimum guarantees (on both RA and SRA products) and that it has been responding by heavily pushing its client institutions to switch away from RA and SRA product lines to other lines with lower guarantees (e.g., RC and RCP with 1% guarantees). If we stay in a low interest rate environment, RA and SRA contract holders with legacy 3% guarantees may end up in the future being a relatively small minority cross-subsidized by other products earning lower rates of return and with different term structures.
Last edited by dodecahedron on Sun Oct 20, 2019 9:06 am, edited 3 times in total.

crefwatch
Posts: 263
Joined: Sun Apr 15, 2007 1:07 pm
Location: New Jersey, USA

Re: Composition and safety of TIAA traditional annuity

Post by crefwatch » Sun Oct 20, 2019 8:41 am

Although it was designed as a comparison of TIAA Traditional and CREF Stock, the chart below may be of general interest for the substantial (if, "past") data it provides on historical returns of TIAA Traditional. It is a scan from "Allocating Premiums between TIAA and CREF", TIAA-CREF, April, 1987, Document number A-5618-487. I don't see a copyright notice on it. I'll also post the following page, with Cumulative figures. These are hosted on my Google Drive, so they may not last forever.

Image

crefwatch
Posts: 263
Joined: Sun Apr 15, 2007 1:07 pm
Location: New Jersey, USA

Re: Composition and safety of TIAA traditional annuity

Post by crefwatch » Sun Oct 20, 2019 8:42 am

Image

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Sun Oct 20, 2019 11:45 pm

dodecahedron wrote:
Sun Oct 20, 2019 8:34 am
Gabriel A. Lozada wrote:
Sat Oct 19, 2019 11:16 pm
In the long run, Traditional cannot sustainably pay more than TIAA's General Account earns. During a period of rising interest rates, the General Account's value will fall and its yield will rise. This will be reflected in Traditional in no raising of its yield initially, and Traditional's eventual yield increases, when they come, will likely lag those of the General Account and of the broader market for a long time.
Your analysis does not reflect the fact that the General Account is not just set up to support the inflows and outflows to the very narrow and specific type of Traditional RA accounts that you analyzed.

The General Account stands behind many different types of TIAA liabilities (including for example, TIAA Life Insurance, both term and cash surrender value life and retained asset accounts, Supplemental Retirement Annuity (SRA) Trad, Retirement Choice (RC) Trad, IRA Trad, just to name a few.) The General Account is also the standby source of liquidity for TIAA Real Estate Annuity (TREA) redemptions. (TREA pays the GA a non-trivial sum for this liquidity guarantee, by the way.) And of course the General Account stands behind annuities in payout mode.

My point is that the operating margins and durations/term structures of all these many very different types of claims against the General Account are very different from the 9 year redemption provision of TIAA RA Trad. It is not appropriate to analyze the performance of the GA as if it only faces a 9 year redemption schedule.

Edited to add: my impression is that TIAA is also quite concerned about the sustainability of their products with 3% minimum guarantees (on both RA and SRA products) and that it has been responding by heavily pushing its client institutions to switch away from RA and SRA product lines to other lines with lower guarantees (e.g., RC and RCP with 1% guarantees). If we stay in a low interest rate environment, RA and SRA contract holders with legacy 3% guarantees may end up in the future being a relatively small minority cross-subsidized by other products earning lower rates of return and with different term structures.
It is quite true that "the General Account stands behind many different types of TIAA liabilities," many of which "are very different from the 9 year redemption provision of TIAA RA Trad." The performance data I use in the paper is for TIAA Traditional, not for the General Account; in Section 4 it comes from Goodman and Richardson (2014), supplemented with sometimes-contradictory more recent data from TIAA, and in Section 3 it comes from Babbel et al. (2015). The role of the General Account in my paper is to determine which combination of Vanguard funds should be used to benchmark Traditional---in particular, to benchmark the Traditional 9-year payout annuity, since, as the paper explains on p. 17, it's not possible to get data on potentially more interesting varieties of Traditional, such as a life annuity. TIAA's liability for 9-year payout annuities is almost certainly shorter than TIAA's life insurance liabilities, and longer than TIAA's liability for versions of Traditional which allow immediate redemptions (and for the Real Estate account which also allows immediate redemptions, as you point out), so the 9-year payout annuities are not extreme outliers in terms of TIAA's liabilities. Granted, that is not saying much; so suppose, if TIAA's only liabilities were 9-year payout annuities, the General Account would be structured differently. If it would be structured with a shorter duration, then instead of the alternatives I used, Total Bond, or a mix of Total Bond and cash, would have been more appropriate benchmarks, and the overall evaluation of Traditional would be more favorable (see Figure 3). On the other hand, if TIAA's only liabilities were 9-year payout annuities and in response the General Account would have been structured with a longer duration, then instead of the alternatives I used, the more appropriate alternatives would have been of even longer duration, and since Figure 3 makes clear that during this period longer duration bonds had overall higher return, in that case the overall evaluation of Traditional would be less favorable.

TIAA could easily solve the difficulty of finding an appropriate benchmark for Traditional by providing one. After all, every mutual fund is legally required to provide an appropriate benchmark. Alas, TIAA doesn't. As I discuss on p. 4, at times TIAA has suggested 10-year Treasuries as a benchmark, but the General Account is not risk-free. At other times, TIAA has suggested no benchmark exists. In my opinion, that is disingenuous. So, wanting to come up with a benchmark and having no reliable one from TIAA, I decided it was best to use the General Account's quality and duration. It's not perfect but it is better than nothing.

It's interesting that you say that "my impression is that TIAA is also quite concerned about the sustainability of their products with 3% minimum guarantees." In contrast, the current version of "TIAA Traditional Annuity: Adding safety and stability to retirement portfolios," available at https://www.tiaa.org/public/pdf/compli ... -paper.pdf, makes an almost identical statement to the 2013 version cited in my paper: "While the 3% guarantee rate exists as a term within the TIAA Traditional Annuity RA and GRA contract, the term does not appear to provide any meaningful value to owners of the contract given the term structure of interest rates and the nature of the yield curve." I am not disputing your impression, just noting what TIAA is still telling the public.

Thank you very much for your comments on the paper!

Trurl Klapaucius
Posts: 53
Joined: Tue Aug 06, 2013 1:11 pm

Re: Composition and safety of TIAA traditional annuity

Post by Trurl Klapaucius » Mon Oct 21, 2019 3:15 am

Gabriel A. Lozada wrote:
Sun Oct 20, 2019 11:45 pm
dodecahedron wrote:
Sun Oct 20, 2019 8:34 am
Gabriel A. Lozada wrote:
Sat Oct 19, 2019 11:16 pm
In the long run, Traditional cannot sustainably pay more than TIAA's General Account earns. During a period of rising interest rates, the General Account's value will fall and its yield will rise. This will be reflected in Traditional in no raising of its yield initially, and Traditional's eventual yield increases, when they come, will likely lag those of the General Account and of the broader market for a long time.
Your analysis does not reflect the fact that the General Account is not just set up to support the inflows and outflows to the very narrow and specific type of Traditional RA accounts that you analyzed.

The General Account stands behind many different types of TIAA liabilities (including for example, TIAA Life Insurance, both term and cash surrender value life and retained asset accounts, Supplemental Retirement Annuity (SRA) Trad, Retirement Choice (RC) Trad, IRA Trad, just to name a few.) The General Account is also the standby source of liquidity for TIAA Real Estate Annuity (TREA) redemptions. (TREA pays the GA a non-trivial sum for this liquidity guarantee, by the way.) And of course the General Account stands behind annuities in payout mode.

My point is that the operating margins and durations/term structures of all these many very different types of claims against the General Account are very different from the 9 year redemption provision of TIAA RA Trad. It is not appropriate to analyze the performance of the GA as if it only faces a 9 year redemption schedule.

...
It is quite true that "the General Account stands behind many different types of TIAA liabilities," many of which "are very different from the 9 year redemption provision of TIAA RA Trad."

...
And, of course, on a basic level, having additional liabilities lowers the overall ability of the General Account to pay claims to Traditional holders.

User avatar
dodecahedron
Posts: 4808
Joined: Tue Nov 12, 2013 12:28 pm

Re: Composition and safety of TIAA traditional annuity

Post by dodecahedron » Mon Oct 21, 2019 5:50 am

Gabriel A. Lozada wrote:
Sun Oct 20, 2019 11:45 pm
Thank you very much for your comments on the paper!
I sent you a Private Message with some further observations.

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Mon Oct 21, 2019 11:51 pm

crefwatch wrote:
Sun Oct 20, 2019 8:41 am
Although it was designed as a comparison of TIAA Traditional and CREF Stock, the chart below may be of general interest for the substantial (if, "past") data it provides on historical returns of TIAA Traditional. It is a scan from "Allocating Premiums between TIAA and CREF", TIAA-CREF, April, 1987, Document number A-5618-487. I don't see a copyright notice on it. I'll also post the following page, with Cumulative figures. These are hosted on my Google Drive, so they may not last forever.
Thank you very much for this data, and for the notice that it "may not last forever." To try to preserve the data even longer, I've tried, I hope accurately, to extract the right-most TIAA columns as text, along with the table headings.

Code: Select all

"start": Start of period, January 1 [of]
"cumul.": Cumulative Compound Rates of Return Over Specified Periods for Each Dollar of Accumulation at Start of Period, TIAA
"annual": Average Annual Compound Rates of Return Over Specified Periods for Each Dollar of Accumulation at Start of Period, TIAA
The "Specified Periods" are "through Dec. 31, 1986."

start cumul.    annual
1953   550.3     5.66
1954   532.9     5.75  
1955   515.9     5.85  
1956   498.0     5.94  
1957   480.6     6.04  
1958   463.0     6.14  
1959   445.3     6.25  
1960   428.2     6.36  
1961   410.3     6.47  
1962   391.9     6.58  
1963   373.5     6.69  
1964   355.3     6.81  
1965   336.7     6.93  
1966   318.9     7.06  
1967   301.9     7.20  
1968   284.6     7.35  
1969   268.0     7.51  
1970   289.4     8.33  
1971   265.1     8.43  
1972   241.2     8.53  
1973   218.9     8.64  
1974   196.9     8.73  
1975   176.2     8.83  
1976   156.9     8.96  
1977   139.0     9.10  
1978   121.9     9.26  
1979   123.9     10.60 
1980   106.6     10.92 
1981   88.9      11.19 
1982   78.4      12.28 
1983   57.0      11.94 
1984   39.5      11.74 
1985   24.1      11.40 
1986   10.3      10.29 
If anyone finds an error in this listing of the data please let me know.

In a remarkable coincidence, the data series I was able to construct started in 1987, the very year the data from crefwatch ends. These are annual rates of return, annually compounded (not continously compounded, as most of the paper's rates of return are). The paper's appendix describes how they were constructed, including a discussion of the uncertainty involved in the numbers. The numbers themselves appear in the paper's last footnote. The paper "may not last forever" in its current place on the web, so I reproduce the data here:

Code: Select all

1987	10.19
1988	9.89
1989	9.61
1990	9.26
1991	8.92
1992	8.30
1993	7.72
1994	7.16
1995	7.10
1996	7.27
1997	7.27
1998	7.27
1999	7.21
2000	7.22
2001	7.37
2002	7.23
2003	6.37
2004	5.68
2005	5.20
2006	5.12
2007	5.48
2008	5.50
2009	4.24
2010	3.94
2011	4.18
2012	4.15
2013	4.31
2014	4.40
2015	4.18

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Tue Oct 22, 2019 1:04 am

Gabriel A. Lozada wrote:
Mon Oct 21, 2019 11:51 pm
[...] I've tried, I hope accurately, to extract the right-most TIAA columns as text [...]
From the cumulative and average annual rates of return for Traditional from crefwatch, I have calculated each year's annual rate of return (annually compounded) for Traditional, given in the table below. For each year, the annual return number calculated from the "cumulative" figures should be the same as that calculated from the "average annual" figures, but they turned out to be slightly different, which is not surprising because, for example, Table 3 gives the 1986 rate of return for Traditional as 10.29 and Table 4 gives it as 10.3.

Code: Select all

        from    from
        cumul.  annual
1953	2.75	2.73
1954	2.76	2.60
1955	2.99	3.10
1956	3.00	2.98
1957	3.13	3.18
1958	3.25	3.11
1959	3.24	3.32
1960	3.51	3.54
1961	3.74	3.76
1962	3.89	3.97
1963	4.00	3.97
1964	4.26	4.20
1965	4.25	4.24
1966	4.23	4.30
1967	4.50	4.39
1968	4.51	4.51
1969	-5.50	-5.52
1970	6.66	6.74
1971	7.00	6.94
1972	6.99	7.00
1973	7.41	7.48
1974	7.49	7.54
1975	7.51	7.41
1976	7.49	7.57
1977	7.71	7.67
1978	-0.89	-0.89
1979	8.37	8.39
1980	9.37	9.31
1981	5.89	5.90
1982	13.63	13.65
1983	12.54	12.54
1984	12.41	12.42
1985	12.51	12.52
1986	10.30	10.29
Notice the slightly negative rate of return for Traditional in 1978 and the very negative rate in 1969! One can confirm this calculation by looking at the cumulative return series (Table 3 of the scans by crefwatch, second-to-last column): if return is positive every year then cumulative return should be falling as one moves forward in time because there are fewer years to go until the ending date of 12/31/1986, but from 1969 to 1970 it rises (from 268.0 to 289.4), and from 1978 to 1979 it rises (from 121.9 to 123.9). No such simple test is available for the "average annual" numbers, but they show the essentially the same negative returns when one extracts the annual figures from them.

Is this a data error by TIAA? Or did Traditional really pay a negative return in 1969 and 1978? If Traditional's customary 3% minimum guaranteed return was in force during those years, then it seems that these historical data sheets scanned by crefwatch contain TIAA data errors. Readers of my paper will know that its Appendix shows modern data errors by TIAA.

I welcome anyone to check the calculations I did in this posting, either by re-doing the calculations from scratch, or by reviewing how I did the calculations by downloading my Mathematica file from http://content.csbs.utah.edu/~lozada/Re ... concise.nb or by downloading a PDF of my Mathematica file from http://content.csbs.utah.edu/~lozada/Re ... oncise.pdf.

Topic Author
Nittany_Lion
Posts: 31
Joined: Fri Jul 20, 2018 10:49 am

Re: Composition and safety of TIAA traditional annuity

Post by Nittany_Lion » Tue Oct 22, 2019 1:20 pm

Hello Gabriel,

OP here. Thanks for posting this very interesting paper.

My thinking on TIAA Traditional is that it might be something of a free lunch if yields continue to drop and one is invested through a liquid GSRA account. Suppose yields continue to fall such that the yield on the weighted combination of VG bonds you considered drops to say 2.00%. If one believes yield is the best predictor of total return going forward, then TIAA would be offering a quality and duration matched product with a superior expected return, no volatility, and no liquidity restrictions for SRA and GSRA products.

In that scenario, the big issue becomes the viability of the 3% guarantee going forward. As you noted, such a situation cannot be sustainable indefinitely, but it seems as though there could be a period of time when one could scoop up this "yield premium," especially if the general account reserves provide a means of meeting the guarantee in low-rate environments. The key would be figuring out when to exit based on the yield disparity and the size of reserves relative to liabilities.

This is less pie-in-the-sky than it seems in that the TIAA real estate account was successfully gamed by many back in 2008 due to the fact that the account's assets were only marked to market once a quarter or so, whereas REITs were crashing and burning (and, indeed, the TIAA real estate account ultimately fell by roughly as much, but with a long and exploitable lag).

I came across a white paper awhile back (link below) that describes what would get them in trouble, albeit in the vague terms you have grown accustomed to. In a nutshell, they will begin to have problems if the ten-year treasury falls below 1.50% for a "significant amount of time" and the spread on corporate bonds (I guess at similar maturities?) falls below 1.48%. They note that historically when the ten-year risk-free rate has fallen, the spread with corporate bonds has tended to rise which, I presume, is what allowed them to meet the guarantee when treasury rates were super low after the financial crisis. This is described in the section near the end titled, "Impact of interest rate fluctuations."

I would very interested in any insight you have to share on what I have described above. And thanks again for your thorough analysis of TIAA Traditional and for helping to pierce the veil on the fund's underlying risk and return characteristics. I hope your paper lands in a good journal.

https://www.tiaa.org/public/pdf/complia ... -paper.pdf

NL

User avatar
Gabriel A. Lozada
Posts: 31
Joined: Wed Oct 16, 2019 10:53 pm
Location: Dept. of Economics, University of Utah, Salt Lake City, Utah, USA 84112
Contact:

Re: Composition and safety of TIAA traditional annuity

Post by Gabriel A. Lozada » Wed Oct 23, 2019 1:17 am

Nittany_Lion wrote:
Tue Oct 22, 2019 1:20 pm
Hello Gabriel,

OP here. Thanks for posting this very interesting paper.

My thinking on TIAA Traditional is that it might be something of a free lunch if yields continue to drop and one is invested through a liquid GSRA account. Suppose yields continue to fall such that the yield on the weighted combination of VG bonds you considered drops to say 2.00%. If one believes yield is the best predictor of total return going forward, then TIAA would be offering a quality and duration matched product with a superior expected return, no volatility, and no liquidity restrictions for SRA and GSRA products.

In that scenario, the big issue becomes the viability of the 3% guarantee going forward. As you noted, such a situation cannot be sustainable indefinitely, but it seems as though there could be a period of time when one could scoop up this "yield premium," especially if the general account reserves provide a means of meeting the guarantee in low-rate environments. The key would be figuring out when to exit based on the yield disparity and the size of reserves relative to liabilities.

This is less pie-in-the-sky than it seems in that the TIAA real estate account was successfully gamed by many back in 2008 due to the fact that the account's assets were only marked to market once a quarter or so, whereas REITs were crashing and burning (and, indeed, the TIAA real estate account ultimately fell by roughly as much, but with a long and exploitable lag).
The scenario you sketch is quite similar to the one conjectured by dodecahedron in a post above:
dodecahedron wrote:
Sun Oct 20, 2019 8:34 am
[...] my impression is that TIAA is also quite concerned about the sustainability of their products with 3% minimum guarantees (on both RA and SRA products) and that it has been responding by heavily pushing its client institutions to switch away from RA and SRA product lines to other lines with lower guarantees (e.g., RC and RCP with 1% guarantees). If we stay in a low interest rate environment, RA and SRA contract holders with legacy 3% guarantees may end up in the future being a relatively small minority cross-subsidized by other products earning lower rates of return and with different term structures.
While I do not think Traditional provided a free lunch during the time period I studied, the future could be different from the past in just the way that you and dodecahedron suggest. In other words, the "Traditional with a 3% guarantee" could offer a free lunch in the future, especially if the general level of interest rates becomes very low, and TIAA would probably respond by restricting access to it and TIAA would have to subsidize it.
Nittany_Lion wrote:
Tue Oct 22, 2019 1:20 pm
I came across a white paper awhile back (link below) that describes what would get them in trouble, albeit in the vague terms you have grown accustomed to. In a nutshell, they will begin to have problems if the ten-year treasury falls below 1.50% for a "significant amount of time" and the spread on corporate bonds (I guess at similar maturities?) falls below 1.48%. They note that historically when the ten-year risk-free rate has fallen, the spread with corporate bonds has tended to rise which, I presume, is what allowed them to meet the guarantee when treasury rates were super low after the financial crisis. This is described in the section near the end titled, "Impact of interest rate fluctuations." [...] https://www.tiaa.org/public/pdf/complia ... -paper.pdf
The white paper you cite, "TIAA Traditional Annuity: Adding safety and stability to retirement portfolios," has been annually updated since at least 2013. In the version dated 8/13, the passages you refer to read as follows:

"We used 23 years of data (1990–2012), the “measurement period,” to determine the blended return on TIAA Traditional Annuity contracts for the participant population. [...] The mean spread over 10-year Treasuries for the observed measurement period is 131 basis points. [...]

"[...] we believe that the long-term risk-free rate would have to remain at 2.5% or less for a significant amount of time for the value of the guaranteed rates on typical retirement annuity contracts, to be worth more than the risk-free rate plus the TIAA credit spread. During periods of low interest rates, we have observed a significant increase in the credit spread. As a result, the potential impact of the prolonged low interest rates has not impacted the valuation of the TIAA Traditional Annuity. As of December 31, 2012, 10-year Treasuries were yielding approximately 1.76%. While the 3% guarantee rate exists as a term within the TIAA Traditional Annuity contract, the term does not appear to provide any meaningful value to owners of the contract given the term structure of interest rates and the nature of the yield curve."

Here is the way the numbers given in red above have changed in some of the more recent versions of this document.
4/15 version: 1990--2014; 136; 2.0%; 2014; 2.17%
9/16 version: 1991--2015; 141; 1.5%; 2015; 2.27%
4/17 version: 1992--2016; 148; 1.5%; 2016; 2.45%
3/18 version: 1993--2017; 150; 1.5%; 2017; 2.40%
3/19 version: 1994--2018; 148; 1.5%; 2018; 2.69%.

TIAA has changed its opinion of how low the 10-year Treasury yield would have to go in order for the 3% guarantee to have "meaningful value." In 2013, TIAA said this level was 2.5% (higher than the actual 10-year Treasury yield on 12/31/12); in 2015, it said this level was 2.0%; and in the years 2016 and later, it said the level was 1.5%. During this time, the backward-looking spread of Traditional over the 10-year Treasury only varied between 131 and 150 basis points, that is, it varied by 19 basis points. If one subtracts 131 or 150 basis points from the 3% guarantee one obtains 1.50% or 1.69%, so those would seem to be the approximate critical yield for the 10-year Treasury below which the 3% guarantee would have to be subsidized by TIAA. This supports TIAA's more recent 1.5% figure. Since April 1953, the 10-year Treasury yield has only fallen below 1.5% briefly during the summer of 2012, briefly during the summer of 2016, and briefly this past summer. It is about 1.8% now.

If, for each of these reports, one sums the 10-year Treasury yield of the preceding Dec. 31 and the reported backward-looking spread of Traditional over the 10-year Treasury, one obtains: 3.07%, 3.53%, 3.68%, 3.93%, 3.90%, and 4.17%. With the possible exception of 2012, this tends to support TIAA's assertion that the 3% guarantee was not meaningful in the past. The future may be different, or not.

None of the Vanguard mutual fund combinations suggested in Table 3 of the paper have a current yield now of over 3%.

I think this exhausts my insight on this matter; going further would require speculating about the future of interest rates, which is hard to do with much accuracy.

Topic Author
Nittany_Lion
Posts: 31
Joined: Fri Jul 20, 2018 10:49 am

Re: Composition and safety of TIAA traditional annuity

Post by Nittany_Lion » Thu Oct 24, 2019 10:08 am

Thank you for taking the time to lay all this out, Gabriel. FWIW, I wasn't disputing your conclusions regarding Traditional's performance over the past 30 years. It accords well with my priors and I really appreciated seeing a detailed analysis documenting that. I believe one can construct a scenario in which there is a free-lunch aspect to the guarantee once the yields on your matched synthetic portfolios no longer outperform Traditional in terms of expected return; which appears to be true now based on your post.

The downside is if rates drop enough to jeopardize the guarantee. It's hard to know what to make of TIAA's claims in that regard, especially since they appear to have moved the goalposts on how low is too low. Alternatively, if one were to believe that TIAA had enough escape hatches via reserves, prohibitions on new investments, etc. to meet the guarantee in virtually all scenarios -- including a 2008-style crisis -- then Traditional might make sense as an alternative, or partial alternative, to a quality and duration matched bond portfolio in a low-rate world. Regrettably, for me at least, your paper provides reason for skepticism in that case as well given the deteroriation in credit quality since 2013.

I am coming around to the view that finding the "sweet spot" on this is more trouble than it's worth and perhaps the best course of action for bond-heavy investors like myself is to view Traditional, circa 2019, as being akin to a barely investment grade corporate bond fund and allocating accordingly.

NL

aristotelian
Posts: 6444
Joined: Wed Jan 11, 2017 8:05 pm

Re: Composition and safety of TIAA traditional annuity

Post by aristotelian » Thu Oct 24, 2019 1:53 pm

Apologies if this is simplistic. I just skimmed the paper and most of it is over my head. However, wouldn't we expect TIAA Traditional to underperform the holdings of the General Account? The General Account has to pay for TIAA Traditional plus at minimum some expenses for TIAA to keep the program going. Of course individual investors would be better off replicating the General Account if they could. Am I missing something?

TSR
Posts: 818
Joined: Thu Apr 19, 2012 9:08 am

Re: Composition and safety of TIAA traditional annuity

Post by TSR » Thu Oct 24, 2019 2:53 pm

What a fascinating thread representing the best of what Bogleheads has to offer. I'm chiming in to say that when my girlfriend was starting her academic job several years ago, she could choose between TIAA and Fidelity, and she was advised by all of her older colleagues to go with TIAA traditional rather than the Fidelity option. They gave her a "just trust me, it's been good to us" kind of explanation. She and I had been steeping ourselves in Bogleheads advice and so I argued that "TIAA can't possibly be that much better than what bond funds can usually offer, so if you can stay the course on bond funds then why not just go with the less opaque investment vehicle?"

My takeaway from this article is that I was right; also, her colleagues were right. That is, this is the rare case where the two options are both pretty good and it's hard to make a strong argument either way. My other takeaway is that there are probably a lot of senior-level professors out there saying the same thing to more junior ones, and there ought to be some better advice than "just trust me." Perhaps that advice is something like, "If you don't want to spend a lot of time thinking about your investments, stick with TIAA for your fixed income. But if you know what you're doing or already have a portfolio with a fixed-income component, you might want to stick with one of Fidelity's low-cost bond funds."

Thanks for the very interesting discussion.

Post Reply