2015 fellows symposium retirement benefits for a …...proportion of respondents with rhi by...
TRANSCRIPT
2015 Fellows Symposium
Retirement Benefits for a 21st Century Workforce
Offering vs. Choice in Retirement Plans:
A Cross Sectional Analysis of Investment
Menus with Traditional and Life-Cycle
Mutual Funds
Tai Kam, Robert L. McDonald,
David P. Richardson and Thomas A. Rietz
October 2, 2015
The General Question:
How do people
allocate their
retirement
assets?
UBS Tactical
Allocation Diagram
Our study
• TIAA-CREF Data:
– Cross section of holdings and allocations
– 654,197 active participants
– 1,073 plans and 2,361 plan combinations
– 98 of the largest institutional clients in 2012
• Questions:
– How do plan characteristics influence
participant decisions?
– How have life-cycle funds affected decisions?
What Should People Do?
• Campbell and Viceira (2002) Figure 7.3: A stylized
lifecycle model of HOLDINGS over time:
Relative allocations correspond
to the relative slopes
CRRA, b=0.96, g=5
rf = 2% RP = 4%
Probabilistic death (max 100)
Deterministic retirement (65)
Mandatory Soc. Sec. savings
10%, invested at 2%
Correlation between stock returns
and labor income 0.3-0.5
What is a Life-Cycle Fund?
https://www.tiaa-cref.org/public/products-services/mutual-funds/lifecycle
Offerings versus Choices
-
100
200
300
400
500
600
1 5 9 131721252933374145495357616569737781
Nu
mb
er o
f P
lans
and T
housa
nds
of
Par
tici
pan
ts
Number of Fund Choices Offered by Plan or Plan
Combination
Panel A: Choices at Plan Level
# of Plans # of Plan Participants (x100)
0%
20%
40%
60%
80%
100%
1 2 3 4 5 6 7 8 9 10 > 10Per
centa
ge
of
Par
tici
pan
ts
Number of Funds Chosen
All Participants No Life-Cycle Funds
Some Life-Cycle Funds Only Life-Cycle Funds
• Participants have lots of choices
• Range: 1 to 84
• Median: 36
• Participants only choose a few
• Median: 2 to 3
Life-Cycle investors are different
Table 2
No Life
Cycle Funds
Some Life
Cycle Funds
Only Life
Cycle Funds
Number of funds 3 4 >> 1
% Equity 60% 64% << 80%
Age 52.13 48.31 >> 40.50
Tenure 15.94 12.86 >> 5.45
Match Rate 187% 193% >> 170%
% Female 52% 54% << 58%
Ann. Contribution $8,880 $9.310 >> $3,630
Compensation $63,400 $61,000 >> $38,800
Plan Assets $115,200 $86,000 >> $13,000
Zip+4 Wealth $156,700 $130,000 >> $57,000
Do Participants Naïvely Diversify? NO
• “1/N” rule:
– Put 1/Nth of allocation into all N funds available
– Result: NO. Same as Huberman and Jiang, JF, 2006
• “Conditional 1/n” rule:
– Put 1/nth of allocation into n<N funds chosen
– Result: NO. Differs from Huberman and Jiang
Are there Menu Effects? YES
• “Number Available Effect”
– Does number of funds available affect number chosen?
– Result: YES (small). Differs from Huberman and Jiang.
• “Equity Exposure Effect” – Does fraction of equity funds offered affect equity allocation?
– Result: YES (economically meaningful). Differs from
Huberman and Jiang.
Participants to use life-cycle funds only
have higher equity allocations…but they
drop faster
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
25-3030-3535-4040-4545-5050-5555-6060-6565-70 70+
% A
llo
cate
d t
o E
qu
ity
Age Range
Female, No Life-Cycle Funds
Male, No Life-Cycle Funds
Female, Some Life-Cycle Funds
Male, Some Life-Cycle Funds
Female, Only Life-Cycle Funds
Male, Only Life-Cycle Funds
Figure 10
Equity exposure and life cycle fund use
affect equity allocations
Variable Overall
Constant 0.5534**
Life-Cycle Only (1/0) 0.4954**
% Equity Offered 0.4755**
Life-Cycle Only x % Equity Offered -0.3770**
Female (1/0) -0.0554**
Life-Cycle Only x Female 0.0486**
Age -0.0056**
Life-Cycle Only x Age -0.0017**
Age x Female 0.0001
Life-Cycle Only x Age x Female 0.0000
Final sample size 645,082
Pseudo R2 20.12%
Table 6
The Problem with Life-Cycle
Funds? • Is the “glide path” optimal for anyone?
• Does one size fit all?
“Households who expect high future labor
income…should have the strongest desire to hold stocks.”
Campbell and Viceira
• What about expenses?
– Retail investors indirectly get institutional fees on
acquired funds, but pay up for the life-cycle fund. Is this
cheaper net than buying funds directly?
2030 Life Cycle Fund Direct and Indirect Expenses
Retail Ret. Premier Inst.
Direct Fund Expenses 0.43% 0.45% 0.30% 0.15%
Acquired Fund Fees and Expenses 0.36% 0.36% 0.36% 0.36%
Reimbursement (runs out 9/30/16) (0.18%) (0.20%) (0.15%) (0.15%)
Total: 0.61% 0.61% 0.51% 0.36%
RHI In Higher Education:
Impact on Retirement
Decisions Robert Clark
TIAA-CREF Fellows Symposium
October 2, 2015
What is Retiree Health
Insurance? Employee benefit that allows retirees to
remain in the employer provided health
plan
Retirees usually required to enroll in
Medicare at age 65 and Medicare
become primary payer
Employer usually subsidizes retirees by
part or all of the premium for the retiree
and in some cases, spouses
Retiree Health Insurance Policies
Retirees still pay premiums, deductibles, and co-pays
Over time, employers alter deductibles and co-pays but tend to retain the percentage subsidy of the premium
Percent of premium paid by employer often related to years of service
A Dying Benefit in Private Sector
Private sector employers have been rapidly terminating RHI plans
Employer has limited control over costs: health care inflation, changes in Medicare, increase in ratio of retirees to workers
Change in accounting rules on reporting in 1990 required reporting unfunded liabilities associated with these plans
Public Employers and
Higher Education
RHI plans still a very common benefit for public employees and for employees of colleges and universities
Trend toward reduced subsidies especially for shorter term employees
Impact of RHI Coverage on
Retirement Decisions
RHI coverage has value
Value is much greater for those
expecting to retire prior to age
65 and Medicare coverage
Employers trying to encourage
workers to retire at younger ages
can couple RHI with generous
pension plans
Retirement Decisions
Affected by RHI
RHI subsidized by the employer implies health expenditures in retirement will lower and individuals can retire at younger ages
RHI subsidized by the employer suggests that individuals need to save less while working to attain the same level of income in retirement
Annual Cost of RHI and
Unfunded Liabilities
With rapidly rising annual
expenditures for RHI, employers
have become increasing
concerned about this benefit
Employers typically do not pre-
fund this promise
Annual Cost of RHI and
Unfunded Liabilities
Thus, many employers now
have large unfunded liabilities
associated with these
programs
State and local governments
now must report unfunded
liabilities of RHI plans
RHI and Higher Education
RHI common in higher education
Academic leaders worried about
the cost of this benefit and
consider options to reduce or
eliminate the cost
Academic leaders worried that the
elimination of RHI could result in
faculty delaying retirement until
even later ages.
Retirement Plans of University
Faculty: Role of RHI
The TIAA-CREF Institute awarded a grant
to fund a survey of older faculty and to
examine how RHI coverage might affect retirement plans
Survey conducted by Mathew
Greenwald & Associates
Survey of approximately 900 full-time
faculty and administrators age 50 and
older employed at U.S. universities
Objectives of Project
Are older faculty covered by RHI planning to retire earlier than those that do not have the ability to remain in the institution’s health plan?
Are older faculty covered by RHI saving less because of the promise of health insurance when they retire?
Economic Literature
There is a short literature examining the
role of RHI on retirement age and
retirement saving but little focus on
public employees and on university
faculty
Two observations:
Public employees tend to retire from
their career jobs at relatively young ages
University faculty tend to retire from their
institutions are relatively older ages
Survey Means: In Percent
Sample With RHI Coverage
Total Sample 48.8
Full Professor 48.4 49.8
Associate Professor 25.7 51.1
Assistant Professor 9.0 42.5
Instructor/Lecturer 9.5 37.6
Administrator 5.4 56.2
Survey Means: In Percent
Sample With RHI Coverage
Public Institution 66.1 57.1
Private Institution 32.5 32.1
Doctoral 57.0 53.7
Masters 25.0 44.4
Baccalaureate 16.5 39.5
Expected Retirement Age
Expected Retirement Age
Without RHI 68.0
Age 50-59 66.1
Age 60 + 70.0
With RHI 68.5
Age 50-59 66.9
Age 60+ 69.7
Expected Retirement Age:
Premium Subsidy
Percent Premium Paid by Faculty Expected Retirement
Age
Zero 67.9
1.0-25.0 68.9
26.0-50.0 67.9
51.0-75.0 67.5
76.0-99.0 70.4
100.0 69.3
RHI and Pension Plans
Type of Pension Percent with RHI Percent without RHI
Covered by DB 62.9 37.1
Covered by DC 41.9 58.1
Regression Results: Expected
Retirement Age
1. Faculty age 50-59 who report that they are in poor health expect to retire almost two years earlier than those in excellent or good health
2. Faculty at doctoral institutions expect to retire at older ages
3. Administrators expect to retire about one year earlier than faculty
4. Faculty at public institutions expect to retire almost one year earlier than those at private institutions
Regression Results: Expected
Retirement Age
5. Males expect to retire about 1.5 years later than females, result is larger among the younger sample
7. Faculty with more years of tenure expect to retire at younger ages
8. Faculty age 50-59 covered by a DB plan expect to retire 2.4 years sooner than those with other types of plans
9. Faculty age 50 to 59 who have developed a retirement plan are expecting to retire 2.5 years sooner than those that have not
Regression Results: Expected
Retirement Age
Key finding for this project:
Coverage by RHI has no effect on
expected retirement age of faculty 60
and over
Coverage of one’s spouse has no effect
on expected retirement age of faculty
Retirement Saving
Most institutions allow faculty to enroll in voluntary supplemental retirement saving plans in addition to their primary pension plan.
These plans can be 401(k), 403(b), or 457 plans
Workers without RHI coverage might be expected to be more likely to enroll in these plans and to have higher contribution rates
Proportion of Respondents with RHI Employed at
Institutions with Voluntary Retirement Savings Plans
Institution Has Supplemental Retirement Plan Sample Mean Has RHI No RHI
Does Not Participate in Plan
Participate in Plan
27.4%
72.6%
54.1%
49.5%
45.9%
50.5%
Proportion of Respondents with RHI by Supplemental Account Balance
Supplemental Account Balance Sample Mean Has RHI No RHI Total
Less than $50,000
12.1% 52.2%
47.8%
100%
$50,000 to less than $100,000 12.3% 58.8%
41.2%
100%
$100,000 to less than $250,000
16.8% 52.7%
47.3%
100%
$250,000 to less than $500,000
14.6% 50.6%
49.4%
100%
$500,000 to less than $750,000
7.2% 57.5%
42.5%
100%
$750,000 to less than $1 million
2.7% 26.7%
73.3%
100%
$1 million or more
3.8% 28.6%
71.4%
100%
Do Not Know 15.7%
49.4%
50.6%
100%
Refused 15.0% 47.0% 53.0% 100%
Participation in
Supplemental Saving Plans
1. Individuals in excellent health are 8.1 percentage
points more likely to participate in a saving plan
relative to those in good health
2. Individuals in bad health are 8.7 percentage point
more likely to participate in a saving plan relative to
those in good health
3. Faculty at public institutions are 7.3 percentage points
less likely to participate in a retirement saving plan
than those at private universities
4. Faculty whose primary plan is a DC plan are less likely
to enroll in a supplemental saving plan
Participation in
Supplemental Saving Plans
Most important for this project
The proportion of Faculty covered
by RHI are not significantly less
likely to enroll in a retirement
saving plan than those that are
not covered by RHI
Key Findings from Project
1. Faculty who report that they anticipate
that their institution will provide them
with health insurance in retirement do
not report that they expect to retire
earlier than those not covered by RHI
2. Faculty who report they anticipate that
their institution will provide them with
health insurance in retirement do not
seem to be saving any less than those
not covered by RHI
LATE-CAREER RESEARCH PRODUCTIVITY AND THE TIMING OF
RETIREMENT
A report on research in progress October 2, 2015
David Blau (Ohio State) John Ham (National U. of Singapore)
Bruce Weinberg (Ohio State)
Supported by a grant from the National Institute on Aging
Overview
• Scientific researchers are typically viewed as being most innovative early in their careers
• The U.S. scientific workforce is aging rapidly
• We don’t know how productivity evolves at the end of the career / near retirement
• The answer to this question is important for understanding the implications of an aging scientific workforce, and possible policy responses
Project Goals 1. Document trends in innovative scientific research output
over the late career
2. Analyze causes of the aging of the scientific workforce, and possible policy responses
3. Estimate effect of retirement on scientific productivity, using variation across universities and within universities over time in discontinuities in retirement incentives induced by Defined Benefit pension plans
4. Simulate impact of the aging of the scientific workforce on the amount of innovative scientific research, and project impact of future trends in aging
Outline of Presentation
• Document aging of the scientific workforce
• Describe pensions and pension incentives
• Next steps
Aging of the U.S. Scientific Workforce
• Use data from the NSF’s Survey of Doctorate Recipients
• Longitudinal survey of representative sample of individuals who received doctorate in U.S.
• Nine waves, 1993-2010
• Restricted-access version of SDR provides extensive career details, and IPEDS code of employer
• Use Census data to compare to aggregate population aging trends
Aging of the U.S. Scientific Workforce
• We have generated full career histories, including years of service with current and former employers, and salary history
• Scientists: – life sciences (biology, medical science, etc.) – physical sciences (chemistry, physics, astronomy, geology) – Engineering – computer science and mathematics – Social science (economics, psychology, etc.). – Not restricted by sector (industry). Educational institutions,
government, private; but pensions available only for public institutions and federal government
The scientific workforce is aging (figure 1); so is the workforce as a whole (figure 2)
0
.01
.02
.03
.04
20 40 60 80Age
scientist_sdr_1993 scientist_sdr_2010
Figure 1: Age Distribution of Scientists 1993 and 2010
0
.01
.02
.03
.04
20 40 60 80Age
workforce_1993 workforce_2010
Figure 2: Age Distribution of Workforce 1993 and 2010
The scientific workforce is older than the workforce as a whole (figure 6), and has been aging more rapidly than the workforce as a
whole (figure 5), especially at older ages (figure 7)
0
.00
2.0
04
.00
6.0
08
20 40 60 80Age
scientist_share_sdr_1993 scientist_share_sdr_2010
Figure 5: Age Distribution of Scientists Share of Workforce 1993 and 2010
42
44
46
48
50
1980 1990 2000 2010 2020Census year
(mean) scientist_mean_age (mean) workforce50_mean_age
Figure 6: Mean Age of Scientists and Workforce
.15
.2.2
5.3
.35
1980 1990 2000 2010 2020Census year
(mean) scientist_sharege55 (mean) workforce50_sharege55
Figure 7: Share of Scientists and Workforce aged 55+
Possible causes of rapid aging of scientific workforce: increase in age at PhD completion
(figure 13), slower rate of retirement (figure 17)
29
30
31
32
33
age
_a
t_ph
d_
sdr
1960 1970 1980 1990 2000 2010Census year
Figure 13: Trend in Age at Science PhD completion
0.1
.2.3
.4
50 55 60 65 70 75Age
1993 rethazard_ 2008 rethazard_
Figure 17: Hazard Rate of Exit to Non-Employment
Other trends of interest: decline in rate of exit from scientific research to other occupations (figure 18), large increase in share of foreign-born among U.S. doctorates
(figure 12)
0
.02
.04
.06
.08
.1
20 30 40 50 60 70Age
1993 sciencehazard 2008 sciencehazard
Figure 18: Hazard Rate of Exit to Non-scientific Employment
.1.2
.3.4
.5
fore
ignb
orn
_sd
r_n
ew
ph
d
1960 1970 1980 1990 2000 2010Census year
Figure 12: Trend in Share Foreign-born in new US Science PhDs
Policy issues • Immigration: government controls who is allowed to enter the US
to work as a scientist, or stay in the US after training
• Subsidies for education and training, to increase number of new scientists
• Science funding policy: increases in NSF and NIH funding could improve career prospects of young scientists by supporting post-doctoral fellowships and increasing success in grant applications from young investigators
• Retirement policy: – Retrospective – impact of elimination of mandatory retirement, and
shift from Defined Benefit to Defined Contribution pension plans
– Prospective – decline in Social Security benefits, financial problems in state pension systems
Defined Benefit Pensions
• We have collected, coded, and are checking DB plans at the top 50 or so research universities with DB plans
– Many of these plans cover multiple state universities
– Current and, to the extent possible, past plan parameters;
changes in plan parameters provide a very useful source of identification
– Buyout windows
– From web sites and plan documents provided by universities
Example of changes in plan: Ohio State. Benefit = b*(Years of Service)*(Final Average Salary)
year Value of b in %
Benefit as % of FAS for retirement with
Gain from 35th year
34 YoS 35 YoS
1965 1.75 59.5 61.25 1.75
1968 1.9 64.6 66.5 1.9
1971 2.0 68 70 2.0
1989 2.1% + 2.5% for each YoS > 30 73 75.5 2.5
1997 2.1% + 2.5% for year 31, 2.6% for year 32,… 73.6 76.5 2.9
1999 2.2% + 2.5% for year 31, 2.6% for year 32,…; + 2.5% instead of 2.2% for each of the first 30 YoS, if YoS >= 35
76.6 88.5 11.9
2015 2.2% 74.8 77 2.2
Illustrate DB pension incentives
• Simulated careers for people turning 65 in 2005 with varying start dates
• Compute “benefit factor” (bf) for alternative retirement ages:
• Benefit = b*YoS*FAS = bf*FAS
• Illustrate increase in benefit factor resulting from delaying retirement by one year
• Use State University of New York (SUNY) DB plan as an example
Illustration of sharp pension incentives
Defined Contribution Pension Plans
• Obtained the 16,175 DC pension plans administered by TIAA-CREF
• We coded them using plan rules in TIAA-CREF data base
• Checked carefully; some ambiguities
• Thanks to Dave and Tai for helping us resolve many of these
• Matched Institution name to IPEDS Institution code, in order to merge with SDR
Match rate to IPEDS
Organization Type # Matched # Unmatched Total Share College (2-Year) 454 132 586 0.775 College (4-Year) 817 243 1060 0.771 University 523 161 684 0.765 Postgraduate 21 11 32 0.656 Teaching Hospital 15 186 201 0.075 Other 256 13,356 13,612 0.016 Total 2086 14,089 16,175 0.129
Illustrate pension Incentives in DC plans
• For each plan, calculate the account balance at each possible age of exit from the employer, using an artificial salary history
• Assume 2% annual real rate of return on balances held in account; no loans or withdrawals before exit
• Hired at age 27; can leave at any age from 28-67
• Converted balance at exit date to a single life annuity using standard life table and 2% interest rate
• Calculated increase in annuity resulting from delaying claiming by one year
• Did this for 91 plans from research universities with “simple” formulas – no discretion over contribution rates
Illustrate for four randomly chosen plans
500
00
100
00
01
50
00
02
00
00
0sa
lary
30 40 50 60 70age
Assumed salary profile
0
200
00
400
00
600
00
800
00
30 40 50 60 70age
1 annuity 2 annuity
3 annuity 4 annuity
Annuity as a function of age at exit from employer
0
200
04
00
06
00
0
30 40 50 60 70age
1 gain 2 gain
3 gain 4 gain
Gain in annuity from delaying exit by one year
0.5
11
.5
30 40 50 60 70age
1 propgain 2 propgain
3 propgain 4 propgain
Proportional gain in annuity from delaying exit by one year
Distribution of proportional gain at ages 50 and 60 across all 91 plans
0.2
.4.6
.8
Fra
ctio
n
.09 .1 .11 .12 .13 .14propgain
Distribution of proportional gain at age 50
0.2
.4.6
.8
Fra
ctio
n
.085 .09 .095 .1 .105propgain
Distribution of proportional gain at age 60
DC Pension Incentives
• Very smooth; no abrupt changes by age or years of service
• There is variation across plans, and some variation within plans over time
• But not as helpful to us in generating exogenous variation in retirement incentives, compared to DB plans
• DB plans have idiosyncratic rules, differ a lot across employers, and sometimes change very abrupt
• Nevertheless, it is important to account for DC pension incentives as control variables
Ongoing work • Quantitative analysis of causes of aging of the scientific workforce
• Use data on publications and patents from SDR to characterize the
age profile of scientific output; currently have self reported counts covering a few years – Awaiting rich data on publication records from NSF
• Project the impact of aging of the scientific workforce on the rate of
scientific output, assuming no changes in behavior – an accounting exercise
• Estimate a model of retirement from the scientific workforce, with pension incentives as key explanatory variables
• Simulate impact of retirement age on scientific output
Age profile of publications per two year period, self-reported
23
45
6
(me
an
) a
rtic
le
30 40 50 60 70 80U_DEM_AGE
2015 Fellows Symposium
Retirement Benefits for a 21st Century Workforce
TIAA-CREF Institute
2015 Fellows Symposium: Retirement Benefits for a 21st Century Workforce
The Performance of TIAA’s Traditional Retirement Annuity 1970-
2013 (David F. Babbel, Mark Meyer, and Miguel Herce)
New York City, October 2, 2015
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
TIAA’s Traditional Retirement Annuity
Data & Summary Statistics
Results
Mean–Variance Analysis
Sharpe & Sortino Ratios
Stochastic Dominance Analysis
Outline
65
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
TIAA’s Traditional Retirement Annuity
Data & Summary Statistics
Results
Mean–Variance Analysis
Sharpe & Sortino Ratios
Stochastic Dominance Analysis
Outline
66
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
TIAA’s Traditional Retirement Annuity (“TIAA RA”) is a deferred fixed annuity, first
introduced in 1918
Flexible Premiums: contributions can be made at regular intervals during the accumulation phase
All premiums contributed during a given period (whose length depends on market conditions)
constitute an investment “vintage”
For each vintage the Company declares annual crediting rates, applying on March 1 of each
year
As time goes by, older vintages tend to merge and receive the same crediting rate
During the decumulation phase, payout options are linked to the participant’s persistence and the
Company’s ability to distribute surplus contingency reserves
We do not study the decumulation phase in this paper
TIAA’s Traditional Retirement Annuity
67
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Contributions received during the period covered by a given investment vintage received
the rates declared for that vintage on March 1 of each year
Crediting rates do not necessarily change every year
As vintages mature they tend to get similar rates
TIAA’s Traditional Retirement Annuity Hypothetical Illustration of the Vintage System of Investment and Rate Crediting
68
Vintage 6 4.50% 4.50% 5.00% 5.00% 3.75% 3.75% 4.00% 4.00% 4.25% 4.25%
Vintage 5 5.00% 5.25% 5.25% 5.00% 5.50% 5.00% 4.75% 4.50% 4.00% 4.25% 4.25%
Vintage 4 6.25% 5.75% 5.00% 5.00% 5.25% 5.25% 4.50% 4.50% 4.00% 4.25% 4.25% 4.25%
Vintage 3 6.75% 6.25% 6.25% 5.75% 5.75% 5.75% 6.00% 5.50% 5.00% 5.00% 4.50% 4.25% 4.25%
Vintage 2 7.25% 6.75% 6.75% 6.25% 6.00% 5.75% 6.00% 6.00% 5.50%% 5.25% 4.75% 4.50% 4.25% 4.25%
Vintage 1 7.75% 8.00% 8.00% 7.25% 6.75% 6.25% 6.25% 6.50% 6.25% 4.25% 4.25% 4.50% 4.25% 4.25% 4.25%
Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 Year 8 Year 9 Year 10 Year 11 Year 12 Year 13 Year 14 Year 15
Note: The crediting rates in this illustration are hypothetical. A participant contributing to all these vintages would
have received six different crediting rates for Year 11
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
TIAA’s Traditional Retirement Annuity – Example Actual Crediting Rates in Effect for the Period 3/1/2014 – 2/28/2015, for Existing Vintages
69
3.50%
3.75%
3.25%
3.75%
4.00%
3.75%
4.00%
4.50%
5.00%
4.25%
4.75%
5.00%
0% 1% 2% 3% 4% 5% 6%
6/1/2014 - 7/31/2014
7/1/2013 - 5/31/2014
1/1/2012 - 6/30/2013
10/1/2011 - 12/31/2011
3/1/2011 - 9/30/2011
9/1/2010 - 2/28/2011
1/1/2010 - 8/31/2010
1/1/2009 - 12/31/2009
1/1/2008 - 12/31/2008
1/1/2000 - 12/31/2007
1/1/1992 - 12/31/1999
Pre-1992
Vin
tage /
Gro
up o
f V
inta
ges
Source: TIAA-CREF, TIAA Retirement Annuity Accounts, TIAA Traditional Annuity, Group Retirement Annuity (GRA), as of 6/30/2014.
By 2014, older
vintages receive
the same
crediting rate
(note vintages
grouped here)
Recent
individual
vintages receive
their own
crediting rates
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
TIAA’s Traditional Retirement Annuity Scope of Our Analysis
70
As we have illustrated, a participant who contributes regular premiums to the TIAA RA will, over time,
participate in a potentially large number of investment vintages
Each vintage will receive its own series of annual crediting rates, applying from March first to
February 28th of each year
The overall annualized return obtained by the participant over a given period of time will be a
blended rate. This is a weighted rate that takes into account the various vintages into which the
participant’s premiums have been pooled
Contrast this with making the same regular contributions into a stock mutual fund, where the full
balance earns a single rate of return over a given year, independently of when the contributions
have been made
In order to compare the performance of the TIAA RA with that of asset classes like stock or bond
mutual funds, we then focus on investment cohorts, where a single investment is made at a point in
time for a given cohort. This is the analysis we conduct in this paper
This approach is the familiar one behind questions like “How much would $1 be worth today if
invested in Berkshire Hathaway on March 1, 1970? And how risky would that be?”
Alternatively, and perhaps more interestingly, the analysis could be: “How would a series of regular
contributions in the TIAA RA have done, compared to the same series invested in an alternative asset?
We do not conduct this analysis in this paper
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
TIAA’s Traditional Retirement Annuity
Data & Summary Statistics
Results
Mean–Variance Analysis
Sharpe & Sortino Ratios
Stochastic Dominance Analysis
Outline
71
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Data and Summary Statistics
72
The TIAA data we use are crediting rates for investment cohorts starting on March 1 of
1970, 1975, 1980, 1985, 1990, 1995, 2000, and 2005
We follow these eight cohorts through December 31, 2013
And compare their separate performance with that of:
Large and small US stocks (“LS” and “SS,” respectively)
Long-term US corporate and government bonds (“LTCB” and “LTGB,” respectively)
Intermediate-term US government bonds (“ITGB”)
3-month U.S. Treasury Bills, a proxy for money market instruments (“MM”)
These data are from Morningstar’s Ibbotson SBBI 2014 Classic Yearbook. Returns are net of
estimated fund fees from ICI
All return series are calculated at the monthly frequency
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Data and Summary Statistics
73
3% Min. Guaranteed
0%
2%
4%
6%
8%
10%
12%
14%
16%
Annualiz
ed M
onth
ly R
etu
rn /
Yie
ld
Annualized Monthly Returns for TIAA RA and 20Y T-Bond Yields
1970 Cohort 1975 Cohort
1980 Cohort 1985 Cohort
1990 Cohort 1995 Cohort
2000 Cohort 2005 Cohort
20Y T-Bond
Sources: TIAA, Ibbotson SBBI 2014 Classic Yearbook
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Cohort Statistic Large Stocks
Small Stocks
Long- Term Corp. Bonds
Long- Term Gov’t Bonds
Interm.- Term Gov’t Bonds
Money Market
TIAA RA
1970 (526)
Mean: 0.79% 1.23% 0.61% 0.59% 0.50% 0.38% 0.58%
STDEV: 4.46% 6.26% 2.80% 3.12% 1.59% 0.26% 0.13%
1975 (466)
Mean: 0.88% 1.40% 0.63% 0.62% 0.51% 0.37% 0.58%
STDEV: 4.36% 5.98% 2.83% 3.19% 1.61% 0.27% 0.14%
1980 (406)
Mean: 0.89% 1.21% 0.72% 0.72% 0.56% 0.35% 0.61%
STDEV: 4.43% 5.79% 2.88% 3.28% 1.61% 0.28% 0.18%
1985 (346)
Mean: 0.85% 1.09% 0.67% 0.69% 0.49% 0.27% 0.58%
STDEV: 4.44% 5.82% 2.56% 3.01% 1.34% 0.20% 0.18%
1990 (286)
Mean: 0.77% 1.18% 0.60% 0.61% 0.44% 0.23% 0.52%
STDEV: 4.28% 5.90% 2.58% 2.92% 1.30% 0.18% 0.12%
1995 (226)
Mean: 0.75% 1.17% 0.58% 0.58% 0.41% 0.20% 0.49%
STDEV: 4.47% 6.24% 2.76% 3.07% 1.28% 0.18% 0.11%
2000 (166)
Mean: 0.36% 0.87% 0.61% 0.56% 0.41% 0.13% 0.49%
STDEV: 4.52% 6.27% 3.00% 3.28% 1.34% 0.16% 0.11%
2005 (106)
Mean: 0.60% 0.90% 0.49% 0.46% 0.35% 0.11% 0.35%
STDEV: 4.43% 6.07% 3.28% 3.48% 1.26% 0.16% 0.04%
Notes: Each cohort’s data span the period from March of the indicated year through December of 2013. The figure in parenthesis under each cohort year is the number of monthly observations in the cohort.
Data and Summary Statistics Sample Means and Standard Deviations for Net Monthly Returns
74
TIAA RA
Average returns
are generally
higher that both
money market
and intermediate-
term government
bond returns, and
their volatility is
much smaller
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
TIAA’s Traditional Retirement Annuity
Data & Summary Statistics
Results
Mean–Variance Analysis
Sharpe & Sortino Ratios
Stochastic Dominance Analysis
Outline
75
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Mean-Variance Analysis Efficient Frontiers – 1970 Cohort (Mar-1970 – Dec-2013)
76
0.00%
0.25%
0.50%
0.75%
1.00%
1.25%
1.50%
1.75%
0% 1% 2% 3% 4% 5% 6% 7%
Ave
rag
e N
et
Mo
nth
ly R
etu
rn
Monthly Standard Deviation
Excluding TIAA RA
Including TIAA RA
Money
Market
TIAA
Intermediate-
Term Gov't Bonds
Long-Term
Government Bonds
Large Stocks
Small Stocks
Long Term
Corporate Bonds
Note the potentially large scope for improvement that inclusion of TIAA RA
investments brings to an optimal mean-variance portfolio for more than two-thirds
of the expected return range.
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Mean Variance Analysis Optimal Portfolio Weights – 1970 Cohort
77
0%
20%
40%
60%
80%
100%
Op
tim
al p
ort
folio
Weig
hts
Monthly Standard Deviation
TIAA RA
Small Stocks
0%
20%
40%
60%
80%
100%
Op
tim
al P
ort
folio
We
igh
ts
Monthly Standard Deviation
Money
Market
Intermediate
Gov't Bonds
Small Stocks
Long Term
Gov't Bonds
Long term
Corporate Bonds
Excluding TIAA
RA Returns
Including TIAA
RA Returns
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Mean Variance Analysis Optimal Portfolio Weights – 1970 Cohort
78
In the absence of the TIAA RA asset, money market instruments, intermediate-term bonds,
and long-term corporate bonds enter the optimal portfolios at various levels of portfolio
risk, along with small stocks and long-term government bonds
Large stocks never enter mean-variance-efficient portfolios model when calibrations are
based on the past 43 years of historical returns and correlations
Including the TIAA RA returns, we observe that no optimal mean-variance portfolio along
the efficient frontier for this cohort includes money market instruments, intermediate-term
bonds or long-term corporate or government bonds. Not even large US stocks are included
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Mean Variance Analysis Optimal Portfolio Weights – 1975 and Subsequent Cohorts
79
In general, our findings for these cohorts are similar to those for the 1970 cohort:
Small stocks are in the portfolios both with and without the TIAA RA investment
Also, the TIAA RA replaces every other asset class except small stocks when it is allowed as an
investment
There are, however, some differences in these cohorts worth noting
When TIAA RA is excluded: long-term corporate bonds now show up in efficient
portfolios for intermediate levels of volatility for the 1980, 1985, 1990, and 2000 cohorts.
The absence of large stocks in any significant level across all cohorts is also quite
remarkable
When TIAA RA is included: the ability of the TIAA RA to replace asset classes other
than small stocks in optimal portfolios is reduced by the significant presence of long-term
government bonds in every cohort. In addition, we observe the appearance of long-term
corporate bonds in the 2000 cohort, and of money market investments for low volatility
levels in the 1985 cohort
These patterns are illustrated with the 1985 cohort:
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Mean Variance Analysis Optimal Portfolio Weights – 1985 Cohort
80
0.00%
0.25%
0.50%
0.75%
1.00%
1.25%
1.50%
1.75%
0% 1% 2% 3% 4% 5% 6% 7%
Avera
ge N
et M
onth
ly R
etu
rn
Monthly Standard Deviation
Excluding TIAA RA
Including TIAA RA
Money
Market
TIAA
Intermediate
Gov't Bonds
Long Term Gov't Bonds
Large Stocks
Small
Stocks
Long Term
Corporate Bonds
0%
20%
40%
60%
80%
100%
Optim
al P
ort
folio
Weig
hts
Monthly Standard Deviation
Money
Market
Small Stocks
Long Term Gov't
Bonds
Intermediate
Gov't Bonds
Long Term
Corporate Bonds
Large Stocks
0%
20%
40%
60%
80%
100%
Optim
al p
ort
folio
Weig
hts
Monthly Standard Deviation
TIAA RA
Small Stocks
Money
Market
Long Term Gov't
Bonds
With no TIAA RA, Long-Term Corporate Bonds now
are in the optimal portfolios
And including TIAA RA, there are also Money Market
and Long-Term Government Bond investments in
some optimal portfolios
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
TIAA’s Traditional Retirement Annuity
Data & Summary Statistics
Results
Mean–Variance Analysis
Sharpe & Sortino Ratios
Stochastic Dominance Analysis
Outline
81
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
R is the asset return, Rf is the risk-free rate of return, E[R – Rf] is the expected value of the
excess of the asset return over the risk-free rate, and Var[R – Rf] is the variance of the
excess return
The Sharpe ratio is used as a measure of how well an investor is compensated per unit of
risk taken. Higher ratios denote greater return for the same level of risk. In this analysis
we take the risk-free rate to be the money market monthly return
The Sortino ratio is based on the Sharpe ratio, but penalizes for only those returns that fall
below the target return, which in our case will be the average riskless rate of return over
the period of analysis
The denominator in the Sortino ratio is the variance formula of the excess returns calculated over
the range of return values where the asset returns of interest are below the risk-free rate
Sharpe and Sortino Ratios
82
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Sharpe and Sortino Ratios Monthly and Annual Returns – 1970 to 1985 Cohorts
83
Large Stocks
Small Stocks
Long-Term Corporate
Bonds Long-Term Gov’t Bonds
Intermediate-Term Gov’t
Bonds TIAA RA
1970 Cohort Monthly Returns
Sharpe Ratio 0.092 0.135 0.080 0.068 0.078 1.075
Sortino Ratio 0.133 0.203 0.125 0.105 0.121 11.737
Annual Returns
Sharpe Ratio 0.281 0.397 0.274 0.259 0.264 1.133
Sortino Ratio 0.467 0.915 0.641 0.656 0.587 9.851
1975 Cohort Monthly Returns
Sharpe Ratio 0.117 0.172 0.090 0.078 0.086 1.080
Sortino Ratio 0.169 0.260 0.140 0.121 0.134 12.808
Annual Returns
Sharpe Ratio 0.353 0.494 0.301 0.293 0.286 1.127
Sortino Ratio 0.600 1.190 0.710 0.763 0.642 10.361
1980 Cohort Monthly Returns
Sharpe Ratio 0.121 0.148 0.129 0.113 0.133 1.554
Sortino Ratio 0.174 0.217 0.209 0.180 0.221 21.761
Annual Returns
Sharpe Ratio 0.358 0.405 0.470 0.457 0.485 1.716
Sortino Ratio 0.613 0.929 1.715 1.792 1.316 16.593
1985 Cohort Monthly Returns
Sharpe Ratio 0.131 0.140 0.155 0.139 0.164 2.461
Sortino Ratio 0.186 0.203 0.251 0.221 0.256 N/A
Annual Returns
Sharpe Ratio 0.382 0.367 0.592 0.582 0.567 2.569
Sortino Ratio 0.654 0.845 2.043 2.765 1.677 N/A
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Sharpe and Sortino Ratios Monthly and Annual Returns – 1990 to 2005 Cohorts
84
Large Stocks
Small Stocks
Long-Term Corporate
Bonds Long-Term Gov’t Bonds
Intermediate-Term Gov’t
Bonds TIAA RA
1990 Cohort Monthly Returns
Sharpe Ratio 0.126 0.161 0.145 0.133 0.167 2.257
Sortino Ratio 0.181 0.243 0.231 0.205 0.260 N/A
Annual Returns
Sharpe Ratio 0.339 0.409 0.645 0.705 0.632 2.326
Sortino Ratio 0.578 0.970 1.909 3.129 1.913 N/A
1995 Cohort Monthly Returns
Sharpe Ratio 0.125 0.155 0.139 0.127 0.165 2.347
Sortino Ratio 0.177 0.234 0.224 0.196 0.263 N/A
Annual Returns
Sharpe Ratio 0.318 0.384 0.630 0.699 0.658 2.407
Sortino Ratio 0.551 0.902 1.945 3.950 2.366 N/A
2000 Cohort Monthly Returns
Sharpe Ratio 0.050 0.118 0.157 0.130 0.210 2.779
Sortino Ratio 0.068 0.173 0.257 0.203 0.342 N/A
Annual Returns
Sharpe Ratio 0.110 0.273 0.787 0.796 0.900 2.778
Sortino Ratio 0.184 0.602 3.285 6.905 4.483 N/A
2005 Cohort Monthly Returns
Sharpe Ratio 0.112 0.130 0.115 0.102 0.193 1.812
Sortino Ratio 0.154 0.188 0.196 0.165 0.333 N/A
Annual Returns
Sharpe Ratio 0.221 0.269 0.550 0.656 0.874 1.708
Sortino Ratio 0.374 0.527 2.338 5.603 16.735 N/A
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
The TIAA RA portfolio has higher Sharpe and Sortino ratios across all the cohorts we
study
For the 1985 through the 2005 cohorts the Sortino ratio is not defined. The reason for this
unusual result is that not a single TIAA RA excess return in these cohorts happens to be
below the corresponding money market return
What we can say from this ratio analysis is that the structure of TIAA RA returns appears
to be very different from that of other asset classes, and that this structure does not lend
itself well to traditional mean-variance metrics for comparison
Moreover, these mean-variance findings are derived from return distributions that, for most
investment classes, are decidedly not normal
Accordingly, we now turn to present alternative and more powerful analyses that buttress
the implications our mean-variance analyses
Sharpe and Sortino Ratios
Conclusions
85
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
TIAA’s Traditional Retirement Annuity
Data & Summary Statistics
Results
Mean–Variance Analysis
Sharpe & Sortino Ratios
Stochastic Dominance Analysis
Outline
86
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Stochastic dominance methods provide no guidance into the construction of a portfolio
from various individual securities, and
Stochastic dominance methods do not yield an equilibrium price for securities.
Stochastic dominance provides evaluative criteria under very general conditions with
respect to an investor’s attitudes toward risk and considers higher moments of the asset
return distributions
The various degrees of stochastic dominance we describe below refer to different aspects
of investors’ attitudes towards risk.
Stochastic Dominance Results
87
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Stochastic Dominance Results
First Degree Stochastic Dominance
88
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
In addition, second degree dominance
(SSD) requires investors to be risk
averse. This implies that a return
distribution that stochastically
dominates another in the second
degree will be preferred by any risk-
averse investor
Graphically, the return distribution F
dominates the return distribution G in
the second degree if:
The distribution F is above the
distribution G for part of the range of
returns,
G starts at a lower return than F, and
The area where the CDF of F is above
the CDF of G is smaller than the area
where the CDF of G is above the CDF
of F
Stochastic Dominance Results
Second Degree Stochastic Dominance
89
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
-20% -10% 0% 10% 20%
Cum
ula
tive P
robabili
ty
Return
Cumulative Distribution G
Cumulative Distribution F
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
In addition, second degree dominance
(SSD) requires investors to be risk
averse. This implies that a return
distribution that stochastically
dominates another in the second
degree will be preferred by any risk-
averse investor
Graphically, the return distribution F
dominates the return distribution G in
the second degree if:
The distribution F is above the
distribution G for part of the range of
returns,
G starts at a lower return than F, and
The area where the CDF of F is above
the CDF of G is smaller than the area
where the CDF of G is above the CDF
of F
Stochastic Dominance Results
Absence of Second Degree Stochastic Dominance
90
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
-20% -10% 0% 10% 20%
Cum
ula
tive P
robabili
ty
Return
Cumulative Distribution G
Cumulative Distribution F
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
The development of third degree stochastic dominance (TSD) was motivated by a long observed
preference among some investors for positively skewed (i.e. asymmetric) returns
A subset of the class of investors who prefer returns exhibiting third degree stochastic dominance is
the important group whose preferences are characterized by decreasing absolute risk aversion
(DARA)
Such investors are willing to pay less for insuring against a given sized risk, on average, as they
accumulate greater wealth, which appears to accord with observed behavior toward risk.
Fourth degree stochastic dominance (4SD) was developed to capture investors’ aversion toward
kurtosis, where returns are characterized by peaked distributions and fat tails, such that losses can be
extreme
Of course kurtosis can favor investors who have asymmetric claims toward returns, such as
investors in call options, but for investors who have equal claims to both tails of a distribution,
such as investors in stocks and bonds, the fatter tails cause a disproportionate loss in utility
Stochastic Dominance Results
Higher Degrees of Stochastic Dominance
91
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Stochastic Dominance Results
1970 Cohort
92
The results shown here are, in most cases, what is to be expected in the sense that it is not common
for empirical return distributions to stochastically dominate other distributions
Remarkably, though, the table shows the TIAA RA return distribution is the dominating one, by the
second degree, over intermediate-term government bonds and money market returns
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Stochastic Dominance Results
1970 Cohort – TIAA RA Dominates Money Market in Second Degree
93
The figure illustrates this result
comparing the TIAA RA return
distribution to the Money Market
return distribution
During the early 1980s, money
market returns were higher than those
provided by the TIAA RA
This happened in 32 cases out of 526
months in the 1970 cohort sample
In the rest of the months, the TIAA
RA returns exceeded those posted by
money market funds
0.0
0.2
0.4
0.6
0.8
1.0
Cum
ula
tive p
robabili
ty
Return
Money Market
TIAA RA
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
Stochastic Dominance Results
Subsequent Cohorts
94
The results we obtain for
the 1970 cohort are
generally the same for
subsequent cohorts
The table to the left shows
the changes with respect to
what is observed for the
1970 cohort
In particular, in addition to
TIAA RA return
distributions dominating
intermediate-term
government bonds in the
second degree,
We observe that they
dominate money market
returns in the first degree
for the 1985 – 2005
cohorts
1975 Cohort Unchanged with respect to the 1970 Cohort
1980 Cohort
Long-term corporate bond returns dominate long-term government bonds in the 2
nd degree
1985 Cohort
Long-term corporate bonds do not dominate long-term government bonds
TIAA RA returns dominates money market returns in the 1
st degree
1990 Cohort
Long-term corporate bonds do not dominate long-term government bonds
TIAA RA returns dominates money market returns in the 1
st degree
1995 Cohort
Long-term corporate bonds do not dominate long-term government bonds
TIAA RA returns dominates money market returns in the 1
st degree
2000 Cohort
Long-term corporate bond returns dominate long-term government bonds in the 2
nd degree
TIAA RA returns dominates money market returns in the 1
st degree
LT corporate and government bonds IT government bonds and the TIAA RA asset class dominate large stocks in the 2
nd
degree
2005 Cohort
TIAA RA returns dominates money market returns in the 1
st degree
Privileged and Confidential
Prepared at the Request of Counsel
Draft: Subject to Revision
Preliminary Work Product
End of Presentation
95
Trends in Retirement Income Distributions by
TIAA-CREF Participants
Jeffrey Brown, Illinois and NBER
James Poterba, MIT and NBER
David P. Richardson, TIAA-CREF Institute
Research Motivation
• Need to understand how retirees draw down assets in a
primary DC environment
• Demographic Pressures
• Aging population
• Increased longevity risk
• Labor Market Shifts
• Majority of workers now covered by DC plans
• Increased labor force by women
• Government Policy Changes
• Social Security and Medicare Reforms
• Qualified plan design reforms
2015 Fall Fellows Symposium: October 2, 2015 97
Research Questions
• What trends do we see from retirees taking distributions from
retirement assets?
• What are the factors that influence the timing, size, and types
of retirement distribution?
• Does plan design play a key role in distribution decisions?
2015 Fall Fellows Symposium: October 2, 2015 98
Prior Research
• King (1996): Trends in the Selection of TC Life Annuity Income
Options, 1978 -1994
• Annuitization required over most of period
• Relatively simple investment menu
• High nominal interest rates
• Women 2x or more likely to choose single-life option
• Women less likely to opt for a guaranteed period
• Trend of average older first annuity age
2015 Fall Fellows Symposium: October 2, 2015 99
Prior Research
• Ameriks (2002): Recent trends in the Selection of Retirement
Income Streams by TIAA-CREF Participants
• Trends over the 1995 – 2001 period
• Full menu of income options
• Simple investment menu
• Falling nominal interest rates
• Continuation of trend of older first annuity age
• Strong proportional take-up of non-annuity options
• Women 2x or more likely to choose single-life option
• Women more likely to opt for a guaranteed period
2015 Fall Fellows Symposium: October 2, 2015 100
Distribution Options
• Income:
• Annuity (1918): fixed or variable, single or joint, guarantee period
option
• Interest Payout Only Option (1989): TIAA only, limited duration
(IPRO)
• Minimum Distribution Option (1991): required minimum distributions
for retirees age 70.5 and older (MDO)
• Systematic Withdrawals and Transfers (1996): participant sets the
schedule (SWAT)
• Other Cash Withdrawals:
• Transfer Payout Annuity (1991): TIAA only, count only cash outs.
(TPA)
• Other lump-sums: periodic or single, cash distributions
2015 Fall Fellows Symposium: October 2, 2015 101
System Snapshot: Distribution Choices of TIAA-CREF
Participants - Age 59.5 or older and non-contributing
2015 Fall Fellows Symposium: October 2, 2015 102
Payment Type% with
income
% with
distribution
% with
assets
Annuity Certain 4% 3% 2%
Joint Life 5% 4% 2%
Joint Life with Guarantee 31% 23% 13%
Single Life 9% 6% 4%
Single Life with Guarantee 17% 13% 8%
Unique Participants with IA Payments 64% 47% 28%
TPA1 13% 8%
IPRO 4% 3% 2%
MDO 34% 25% 15%
SWAT 6% 4% 3%
Cash Withdrawals 21% 12%
Unique Participants with DA Payments2 47% 28%
Unique Participants with Income Payments 74% 44%
Unique Participants with Payments 59%
Unique Participants with No Payments 41%
Total Participants 446,551 604,561 1,019,076
2 Rollover to IRA and plan to plan asset transfers are not included.
1 Include periodic cash disbursements. Rollover to IRA and asset transfers to other CREF accounts are
not included.
Initial Income Selections
Source: author calculations of TIAA-CREF administrative records
2015 Fall Fellows Symposium: October 2, 2015 103
Initial Income Selections
Source: author calculations of TIAA-CREF administrative records
2015 Fall Fellows Symposium: October 2, 2015 104
Who Annuitizes and How?
• The extensive margin: Do I annuitize?
• A “working life”, as opposed to “at retirement”, decision
for many TC participants.
• The intensive margin: How do I annuitize?
• A “during retirement” decision for most participants.
• Financial considerations
• Pricing/interest rate
• Estate planning
• Psychological considerations
• Inertia/Uncertainty
• Mental Accounting
2015 Fall Fellows Symposium: October 2, 2015 105
Asset Allocations, by age cohort
Source: Richardson, David P. and Benjamin Bissette (2014) “Trends in Premium and Asset
Allocations by TIAA-CREF Participants: 2005 – 2011.”TIAA-CREF Institute Research Dialogue
20 20
28
35
42
1117
24
34
44
6770
65
59
52
39
51
55
51
47
77
5 5 5
3
5
4
4
32
2 1 1 1
4
5
4
3
34 2 1 1 0
43
23
138
4
0
10
20
30
40
50
60
70
80
90
100
Under 35 35 - 44 45 - 54 55 - 64 65 + Under 35 35 - 44 45 - 54 55 - 64 65 +
Average Asset Allocations by AgeDecember 2006 and December 2011
TIAA Traditional CREF TIAA Real Estate Retirement Mutual Funds Lifecycle Funds
December 2006 December 2011
2015 Fall Fellows Symposium: October 2, 2015 106
New Settlements: TIAA vs. CREF vs. REA
Source: author calculations of TIAA actuarial data
2015 Fall Fellows Symposium: October 2, 2015 107
Single v. Joint Life, by Gender
Source: author calculations of TIAA-CREF administrative data
2015 Fall Fellows Symposium: October 2, 2015 108
Guarantee period usage, by Gender and annuity type
Source: author calculations of TIAA-CREF data
2015 Fall Fellows Symposium: October 2, 2015 109
First Life Annuity Issue Ages
Source: author calculations of TIAA-CREF administrative data
2015 Fall Fellows Symposium: October 2, 2015 110
Final Thoughts
• Growing popularity of single life annuities and guarantee
periods
• Possible interest rate effect?
• Continuing trend of delaying first annuity draw.
• Counter to optimal Social Security strategy?
• Having annuities in the investment menu during the
accumulation phase seems important.
• “Working life” instead of “at retirement” decision.
• Can help participants overcome behavioral biases.
2015 Fall Fellows Symposium: October 2, 2015 111
How Retirees Manage Retirement
Savings for Retirement Income
Paul J. Yakoboski, Ph.D.
Senior Economist, TIAA-CREF Institute
TIAA-CREF Institute Fellows Symposium
Retirement Benefits for a 21st Century Workforce
October 2, 2015
113
Savings to Income Survey
What factors influence how retirees with significant assets in tax-qualified
retirement accounts convert their savings to income during retirement?
• In particular, what are the similarities and differences between those
who annuitize some of their savings and those who do not?
Surveyed 1,000 retired TIAA-CREF participants
• Age 60 or older
• Retired with at least $400,000 in DC and/or IRA assets
• No DB pension income
• 500 receiving annuitized payments (annuitants)
• 500 not receiving annuitized payments (non-annuitants)
114
Annuitant and non-annuitant demographics
42%
61%
47%
30%
9%
4% 2% 5%
0%
20%
40%
60%
Annuitants Non-annuitants
Current age
60-74 75-84 85 or older DNA
18%
31%
23%
31% 30%
26%
29%
13%
0%
10%
20%
30%
40%
Annuitants Non-annuitants
Years retired
2-4 years 5-9 years
10-15 years over 15 years
115
Retirement outcomes
28%
18%
61%
74%
11% 8%
0%
20%
40%
60%
80%
Annuitants Non-annuitants
How did your standard of living change with retirement?
Increased Did not changeDecreased
31%
19%
64%
78%
5% 2%
0%
20%
40%
60%
80%
Annuitants Non-annuitants
How has your lifestyle in retirement compared with your pre-retirement expectations?
Exceeded Met Fallen short
116
Retirement outcomes
Extremely
confident
Very
confident
Somewhat
confident
Not too/not at
all confident
…be able to maintain a comfortable standard of living throughout
retirement?
Annuitants 19% 54% 26% 2%
Non-annuitants 22 52 23 2
…not outlive your savings and financial assets?
Annuitants 22 49 25 3
Non-annuitants 21 47 27 4
…have the best strategy to manage and draw income from your
retirement savings during retirement?
Annuitants 16 48 32 4
Non-annuitants 14 45 34 6
How confident are you that you (and your spouse) will…
117
Top financial priorities in retirement
Very high High Moderate Low/not
Ensuring the financial security of your spouse
if you die first
Annuitants 57% 36% 5% 3%
Non-annuitants 51 36 8 5
Not outliving savings and financial assets
Annuitants 54 35 7 4
Non-annuitants 49 37 10 4
Having a guaranteed income stream sufficient to cover basic expenses
Annuitants 53 38 6 3
Non-annuitants 36 35 18 10
How much of a priority is [this issue] when it comes to
managing your personal finances during retirement?
118
Low financial priorities in retirement
Very high High Moderate Low/not
Having the flexibility to adjust your income as needed over time
Annuitants 15% 49% 27% 10%
Non-annuitants 25 45 26 4
Earning a high rate of return on your financial assets
Annuitants 9 30 49 12
Non-annuitants 11 31 48 10
Leaving an inheritance
Annuitants 6 26 35 34
Non-annuitants 12 27 32 28
Having professionals manage your financial assets
Annuitants 12 20 24 43
Non-annuitants 10 20 24 46
How much of a priority is [this issue] when it comes to
managing your personal finances during retirement?
119
Mid-level financial priorities in retirement
Very high High Moderate Low/not
Maintaining direct control of your financial assets
Annuitants 30% 42% 19% 8%
Non-annuitants 33 47 17 3
Preserving your financial assets
Annuitants 25 45 24 5
Non-annuitants 32 45 20 3
Maintaining the same standard of living throughout retirement
Annuitants 24 55 19 2
Non-annuitants 27 49 22 2
How much of a priority is [this issue] when it comes to
managing your personal finances during retirement?
120
Annuitants’ motivations
Extremely
important
Very
important
Somewhat
important
Not too/
not at all
important
Cannot outlive the
income stream 50% 34% 10% 7%
Providing income for
spouse if annuitant dies
first
49 32 8 11
The certainty of a
constant level of
income
39 39 16 7
Helping to cover basic
living expenses 37 43 14 5
Maintaining standard of
living over time 28 44 22 6
Reasons for annuitizing retirement savings
121
Non-annuitants’ motivations
Extremely
important
Very
important
Somewhat
important
Not too/
not at all
important
Wanting to keep direct
control of the money 26% 41% 25% 8%
Being unable to access
the money if needed 18 37 26 18
Having Social Security
or other income sources 18 34 30 18
Think it is a poor
investment 14 25 35 26
The possibility of losing
money if you die
prematurely
19 26 24 32
The expense of fees and
charges 13 24 33 31
Reasons for not annuitizing retirement savings
122
More on non-annuitants
3% 7%
16%
40%
35%
0%
10%
20%
30%
40%
Likelihood of annuitizing in the future
28% 25%
48%
0%
10%
20%
30%
40%
50%
A pretty goodidea
Somewhat ofan idea
No idea
Would you have an idea of how much income your savings would provide if annuitized?
123
So what’s going on?
Annuitants 54%
Non-annuitants 58%
Amount of advice typically followed?
All Most Some Little/none
Annuitants 21% 49% 26% 4%
Non-annuitants 17 51 26 7
Advice received about annuitizing retirement savings?
Do Don’t No advice
Annuitants 60% 9% 30%
Non-annuitants 21 37 42
Worked with a financial advisor in deciding how to
manage and draw income from retirement savings?
124
So what’s going on?
Yes 25%
No 48%
Don't know/ Not sure
27%
Did you participate in a plan while working that had a deferred annuity as an investment option?
Non-annuitants
125
Key takeaways
Annuitants are more likely to have experienced an increased standard of living in
retirement and a lifestyle that has exceeded preretirement expectations.
Annuitants and non-annuitants share the same top financial priorities for their
personal finances in retirement, each of which is consistent with annuitization.
Furthermore, the most important reasons for retirees deciding to annuitize are
consistent with these top financial priorities.
So why do some retirees annuitize, while others do not? It is possible that non-
annuitants do not understand that annuitization would address their top financial
priorities.
• 80% of non-annuitants were advised to not annuitize or did not receive
advice regarding annuitization; 60% of annuitants were advised to do so.
• In-plan deferred annuities present an opportunity for participants to
become socialized to annuities and annuitization. For 75% of non-
annuitants, either their plan(s) did not have a deferred annuity or they did
not realize it, the latter likely being the case for many.
2015 Fellows Symposium
Retirement Benefits for a 21st Century Workforce