Inconsistent Retirement Timing

Christoph Merkle; Philipp Schreiber; Martin Weber

doi:10.3368/jhr.0920-11215R2

Article Figures & Data

Figures

Tables
Additional Files

Download figure
Open in new tab
Download powerpoint
Figure 1
Time Preference Survey Questions
Notes: Screenshot of the survey question used to elicit time preferences. Displayed are six choices between a smaller sooner amount and a later larger amount. Time consistency requires the same choice (A or B) on the right and the left.
Download figure
Open in new tab
Download powerpoint
Figure 2
Average Actual Retirement Age by Number of Inconsistent Answers
Notes: The bars show the average actual retirement age for participants with zero, one, two, or three inconsistent answers. Significance: *p < 0.1, **p < 0.05, ***p < 0.01, relative to the group with zero inconsistent answers.
Download figure
Open in new tab
Download powerpoint
Figure 3
Fraction of Participants Stating They Retired Too Early
Notes: The bars show the fraction of participants who state they retired too early for participants with zero, one, two, or three inconsistent answers. Significance: *p < 0.1, **p < 0.05, ***p < 0.01, relative to the group with zero inconsistent answers.
Download figure
Open in new tab
Download powerpoint
Figure 4
Planned Retirement Age by Age Group and Time Preferences
Notes: The figure shows the average planned retirement age depending on current age for time-consistent participants (solid line) and for time-inconsistent participants (dashed line).
Download figure
Open in new tab
Download powerpoint
Figure 5
Retirement Timing by Overdraft Usage and Smoking Habits
Notes: The figure shows the fraction of participants who retire earlier than planned in the 2008–2011 SAVE waves. Earlier than planned retirement is assumed if the planned retirement age in wave t – 1 (PRA_t_–1) is larger than the actual retirement age indicated in wave t (ARA_t). Significance: *p < 0.1, **p < 0.05, ***p < 0.01, relative to the group of nonsmokers or to the group not using overdrafts (overdraft = 1).

View popup

Table 1

Detailed Summary Statistics for Time Preferences

Panel A	Retirees			Nonretirees
	Sooner	Later	% Sooner	Sooner	Later	% Sooner
Immediate tax refund (low interest rate)	154	39	80	1,723	409	81
Immediate tax refund (medium interest rate)	112	81	58	946	1,186	44
Immediate tax refund (high interest rate)	104	89	54	535	1,597	25
Delayed tax refund (low interest rate)	116	77	60	1,278	854	60
Delayed tax refund (medium interest rate)	73	120	38	393	1,739	18
Delayed tax refund (high interest rate)	79	114	41	157	1,975	7
Panel B		Retirees			Nonretirees
Observations Included in Analysis		n	%		n	%
Time consistent
# inconsistent answers = 0		122	63.2		1,257	59.0
Present biased
# inconsistent answers = 1		38	19.7		490	23.0
# inconsistent answers = 2		13	6.7		185	8.7
# inconsistent answers = 3		14	7.3		179	8.4
Total		187	96.9		2,111	99.0
Observations excluded from analysis		n	%		n	%
Future biased		4	2.1		17	0.8
Mixed biased		2	1.0		4	0.2
Total		6	3.1		21	1.0

Notes: The table presents detailed summary statistics of the time preference measure used in the survey (see Figure 1). Panel A shows choices of participants (sooner or later tax refund) in each of the six questions separately for the sub-samples of retirees and nonretirees. Panel B specifies time-consistent responses, present-biased responses, and other response patterns.

View popup

Table 2

Summary Statistics

	Retirees (n = 187)		Nonretirees (n = 2,111)
	Mean	SD	Mean	SD
Panel A: Retirement Timing
Planned retirement age			64.97	3.81
Actual retirement age	61.54	4.64
Full retirement age	65.14	0.25	66.72	0.52
Retirement regret	0.22	0.41
Panel B: Time Preferences
Inconsistency indicator	0.35	0.48	0.40	0.49
Number inconsistent answers [0–3]	0.57	0.92	0.66	0.95
Impatience [1–7]	3.85	1.77	3.88	1.62
Panel C: Demographics
Age	66.66	5.70	40.10	12.10
Gender [male = 1]	0.88	0.33	0.85	0.36
Income			3,411.73	3,125.81
Retirement benefits	3,157.41	4,639.30
Satisfaction with benefits [1–7]	5.00	1.83
Number of children	1.64	1.19	0.78	1.19
Education [0–2]	1.44	0.78	1.61	0.62
Married	0.82	0.39	0.47	0.50
Panel D: Additional Controls
Risk aversion [1–7]	4.02	1.51	3.89	1.46
Loss aversion [1–7]	4.67	1.59	4.22	1.59
Financial literacy [0–6]	3.98	0.93	4.15	1.13
Subjective life expectancy	84.35	6.45	83.53	7.78
Private pension insurance	0.36	0.48	0.65	0.48

Notes: The table shows summary statistics for the FAZ survey separately for the subsamples of retired and nonretired participants. Variables are as defined in the Appendix.

View popup

Table 3

Effect of Time Preferences on Actual Retirement Age

	Actual Retirement Age
	(1)	(2)	(3)	(4)	(5)
Inconsistency indicator	−1.758
	(0.80)^**
Number inconsistent answers		−1.142	−1.028	−1.026	−0.978
		(0.46)^**	(0.43)^**	(0.41)^**	(0.41)^**
Impatience			−0.330	−0.443	−0.538
			(0.20)^*	(0.19)^**	(0.21)^**
Age				0.321	0.301
				(0.07)^***	(0.07)^***
Gender				1.507	2.305
				(1.29)	(1.35)^*
Married				−0.348	−0.781
				(0.90)	(1.02)
Number of children				−0.247	−0.311
				(0.28)	(0.30)
Education				0.851	0.681
				(0.48)^*	(0.51)
Retirement benefits [log]				−0.556	−0.458
				(0.30)^*	(0.27)^*
Satisfaction retirement benefits				−0.070	−0.039
				(0.17)	(0.18)
Risk aversion					−0.349
					(0.25)
Loss aversion					−0.026
					(0.23)
Financial literacy					−0.980
					(0.41)^**
Life expectancy					−0.022
					(0.05)
Private pension insurance					0.113
					(0.60)
Constant	62.158	62.195	63.401	45.331	53.353
	(0.34)^***	(0.34)^***	(0.76)^***	(4.66)^***	(7.72)^***
R²	0.028	0.046	0.056	0.250	0.278
Observations	185	185	185	169	160

Notes: The table shows the results of five OLS regressions with actual retirement age in years as the dependent variable. Independent variables are as defined in the Appendix. Robust standard errors are displayed in parentheses. Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table 4

Ex Post Evaluation of the Retirement Decision

	Retirement Regret
	(1)	(2)	(3)	(4)	(5)
Inconsistency indicator	0.188
	(0.07)^***
Number inconsistent answers		0.097	0.097	0.097	0.079
		(0.04)^***	(0.04)^**	(0.04)^**	(0.04)^**
Impatience			0.001	−0.005	−0.006
			(0.02)	(0.02)	(0.02)
Actual retirement age				0.000	−0.002
				(0.01)	(0.01)
Age				−0.001	−0.000
				(0.01)	(0.01)
Gender				−0.100	−0.061
				(0.13)	(0.14)
Married				−0.037	−0.088
				(0.10)	(0.11)
Number of children				0.004	0.000
				(0.03)	(0.03)
Education				−0.018	0.016
				(0.04)	(0.04)
Retirement benefits [log]				0.030	0.025
				(0.02)	(0.02)
Satisfaction retirement benefits				−0.059	−0.072
				(0.02)^***	(0.02)^***
Risk aversion					0.019
					(0.02)
Loss aversion					0.002
					(0.02)
Financial literacy					−0.046
					(0.04)
Life expectancy					0.003
					(0.00)
Private pension insurance					−0.048
					(0.07)
Constant	0.150	0.161	0.156	0.412	0.506
	(0.03)^***	(0.03)^***	(0.07)^**	(0.57)	(0.74)
R²	0.043	0.042	0.036	0.082	0.099
Observations	185	185	185	167	158

Notes: The table shows the results of five linear probability models (OLS) with the indicator retirement regret as the dependent variable. Robust standard errors are displayed in parentheses. Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table 5

Effect of Time Preferences on Planned Retirement Age

	Planned Retirement Age
	(1)	(2)	(3)	(4)	(5)	(6)
Inconsistency indicator	−0.365	−0.327			−0.141	−0.206
	(0.17)^**	(0.17)^*			(0.16)	(0.16)
Impatience		−0.172		−0.172	−0.181	−0.158
		(0.05)^***		(0.05)^***	(0.05)^***	(0.05)^***
Number inconsistent answers			−0.236	−0.221
			(0.08)^***	(0.08)^***
Age					−0.053	−0.026
					(0.01)^***	(0.01)^*
Age × Incon. indicator					−0.034	−0.032
					(0.01)^***	(0.01)^**
Gender						−0.038
						(0.23)
Married						−0.395
						(0.21)^*
Number of children						0.051
						(0.14)
Education						0.495
						(0.13)^***
Income [log]						−0.447
						(0.13)^***
Risk aversion						0.091
						(0.08)
Loss aversion						−0.127
						(0.07)^*
Financial literacy						−0.122
						(0.08)
Life expectancy						0.042
						(0.02)^**
Private pension insurance						−0.552
						(0.17)^***
Constant	65.114	65.763	65.123	65.780	67.862	67.155
	(0.11)^***	(0.22)^***	(0.10)^***	(0.22)^***	(0.45)^***	(2.19)^***
R²	0.002	0.006	0.003	0.008	0.057	0.074
Observations	2,111	2,104	2,111	2,104	2,104	2,046

Notes: The table shows the results of six OLS regressions with planned retirement age as the dependent variable. All variables are as defined in the Appendix. Age × Incon. indicator is the interaction between age and the inconsistency indicator. Robust standard errors are displayed in parentheses. Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table 6

Effect of Time Preferences on Planned Retirement Age—Sample Split

	Planned Retirement Age
	Age ≤ Median		Age > Median		Age > 50	Age > 55
	(1)	(2)	(3)	(4)	(5)	(6)
Number inconsistent answers	0.201	0.125	−0.458	−0.446	−0.774	−0.988
	(0.12)^*	(0.12)	(0.14)^***	(0.14)^***	(0.18)^***	(0.29)^***
Impatience	−0.286	−0.246	−0.124	−0.127	−0.114	−0.243
	(0.07)^***	(0.07)^***	(0.08)	(0.08)	(0.10)	(0.15)
Age		0.021		−0.011	−0.049	−0.175
		(0.04)		(0.02)	(0.06)	(0.20)
Gender		0.054		−0.288	−0.808	−1.022
		(0.31)		(0.36)	(0.54)	(0.73)
Married		0.244		−0.739	−0.747	−0.494
		(0.41)		(0.28)^***	(0.39)^*	(0.55)
Number of children		−0.325		0.239	0.085	0.259
		(0.46)		(0.08)^***	(0.16)	(0.21)
Education		0.827		0.225	0.700	0.548
		(0.25)^***		(0.16)	(0.21)^***	(0.29)^*
Income [log]		−0.848		−0.181	−0.159	0.223
		(0.20)^***		(0.17)	(0.22)	(0.22)
Risk aversion		0.096		0.086	−0.179	−0.258
		(0.11)		(0.10)	(0.15)	(0.19)
Loss aversion		−0.225		−0.039	0.008	0.198
		(0.11)^**		(0.09)	(0.13)	(0.17)
Financial literacy		−0.151		−0.113	−0.049	0.274
		(0.11)		(0.14)	(0.19)	(0.28)
Life expectancy		0.016		0.076	0.046	0.089
		(0.03)		(0.02)^***	(0.03)	(0.04)^**
Private pension insurance		−0.737		−0.441	−0.553	−0.614
		(0.24)^***		(0.28)	(0.38)	(0.54)
Constant	66.810	71.423	64.881	61.319	66.565	65.873
	(0.31)^***	(2.80)^***	(0.32)^***	(2.62)^***	(4.69)^***	(12.28)^***
R²	0.014	0.058	0.013	0.041	0.074	0.139
Observations	1,071	1,043	922	897	375	151

Notes: The table shows six OLS regressions with planned retirement age as the dependent variable. The columns show results for different subsamples by age (participants aged 60 or above are excluded). All variables are as defined in the Appendix. Robust standard errors are displayed in parentheses. Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table 7

Effect of Time Preferences on Owning Private Pension Insurance

	Private Pension Insurance Indicator
	(1)	(2)	(3)	(4)	(5)
Inconsistency indicator	−0.045
	(0.02)^**
Number inconsistent answers		−0.020	−0.020	−0.025	−0.025
		(0.01)^*	(0.01)^*	(0.01)^**	(0.01)^**
Impatience			−0.007	−0.005	−0.005
			(0.01)	(0.01)	(0.01)
Age				−0.000	−0.000
				(0.00)	(0.00)
Gender				−0.004	−0.016
				(0.03)	(0.03)
Married				0.063	0.062
				(0.03)^**	(0.03)^**
Number of children				−0.007	−0.010
				(0.01)	(0.01)
Education				0.018	0.016
				(0.02)	(0.02)
Income [log]				0.069	0.068
				(0.01)^***	(0.01)^***
Risk aversion					−0.019
					(0.01)^**
Loss aversion					−0.005
					(0.01)
Financial literacy					−0.006
					(0.01)
Life expectancy					0.003
					(0.00)^**
Constant	0.649	0.644	0.671	0.099	0.002
	(0.01)^***	(0.01)^***	(0.03)^***	(0.10)	(0.17)
R²	0.002	0.001	0.001	0.025	0.030
Observations	2,277	2,277	2,270	2,072	2,046

Notes: The table shows the results of five linear probability models with the indicator private pension insurance as the dependent variable. All variables are as defined in the Appendix. Robust standard errors are displayed in parentheses. Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table 8

Summary Statistics SAVE Survey

	Retirees (n = 907)		Nonretirees (n = 1,140)
	Mean	SD	Mean	SD
Panel A: Retirement Timing
Planned retirement age			64.92	2.98
Actual retirement age	58.90	6.82
Panel B: Time Preferences
Smoker [0–1]	0.14	0.35	0.31	0.46
Overdraft [1–4]	1.69	1.03	2.15	1.14
Overdraft indicator [0–1]	0.18	0.39	0.32	0.47
Panel C: Demographics
Age	69.83	8.85	46.21	19.26
Gender [male = 1]	0.52	0.50	0.45	0.50
Income (if >0)	1,517.12	1,108.73	1,519.73	985.16
Number of children	2.04	1.41	1.68	1.37
Education [0–2]	0.45	0.79	0.50	0.78
Married	0.61	0.49	0.63	0.48
Panel D: Additional Controls
Financial literacy [0–9]	2.94	1.08	2.95	1.04
Subjective life expectancy	80.33	7.06	78.60	8.40
Private pension insurance (PPI)	0.08	0.28	0.44	0.50
Panel E: Health Controls
Health status [1–5]	3.16	0.83	3.58	0.80
Satisfaction health [0–10]	5.54	2.45	6.44	2.31
Prolonged illness	0.69	0.46	0.43	0.50

Notes: The table presents summary statistics for the SAVE 2010 survey for subsamples of retirees and nonretirees. Variables are as defined in the Appendix.

View popup

Table 9

Actual Retirement Age SAVE Survey

	Actual Retirement Age
	(1)	(2)	(3)	(4)	(5)	(6)
Smoker [0/1]	−3.605	−3.548	−2.661
	(0.79)^***	(0.79)^***	(0.78)^***
Overdraft [1–4]				−1.679	−1.589	−1.394
				(0.27)^***	(0.27)^***	(0.25)^***
Health status		1.577	1.327		1.428	1.369
		(0.48)^***	(0.47)^***		(0.49)^***	(0.47)^***
Satisfaction health		0.017	0.014		0.073	−0.017
		(0.15)	(0.14)		(0.15)	(0.15)
Prolonged illness		0.397	0.658		0.423	0.422
		(0.57)	(0.55)		(0.58)	(0.54)
Gender			1.388			1.053
			(0.48)^***			(0.47)^**
Married			−0.689			−1.345
			(0.5)			(0.5)^***
Number of children			0.103			0.197
			(0.16)			(0.16)
Education			0.874			0.725
			(0.24)^***			(0.26)^***
Income [log]			−0.269			−0.216
			(0.06)^***			(0.06)^***
Financial literacy			−0.104			−0.112
			(0.2)			(0.2)
Life expectancy			0.153			0.142
			(0.03)^***			(0.04)^***
Private pension insurance (PPI)			−4.506			−4.14
			(1.07)^***			(1.01)^***
Constant	59.395	54.042	42.748	61.843	56.48	46.675
	(0.23)^***	(1.53)^***	(3.01)^***	(0.44)^***	(1.59)^***	(3.38)^***
R²	0.033	0.062	0.150	0.067	0.097	0.208
Observations	907	907	905	753	753	752

Notes: The table shows the results of six cross-sectional OLS regressions for the 2010 wave of the SAVE survey. The dependent variable is actual retirement age using the subsample of retired participants. Regressions include either smoking or overdraft usage as a proxy for time inconsistency. Variables are as defined in the Appendix. All five imputations of the SAVE data are used. Coefficients and standard errors are calculated according to Rubin (1987). Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table 10

Planned Retirement Age SAVE Survey

Panel A	Full Panel	Smoker			Nonsmoker
		All	Age > 40	Age > 50	All	Age > 40	Age > 50
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
Wave	−0.093	−0.197	−0.256	−0.321	−0.036	−0.055	−0.085
	(0.05)^*	(0.09)^**	(0.12)^**	(0.15)^**	(0.05)	(0.06)	(0.09)
Health status	0.010	−0.011	0.006	0.046	−0.012	−0.014	0.125
	(0.03)	(0.06)	(0.06)	(0.08)	(0.04)	(0.05)	(0.05)^**
Satisf. health	−0.100	0.028	−0.003	−0.074	−0.165	−0.131	−0.094
	(0.14)	(0.27)	(0.28)	(0.34)	(0.15)	(0.14)	(0.18)
Prolonged illness	0.083	−0.067	−0.198	−0.403	0.19	0.122	−0.073
	(0.13)	(0.25)	(0.32)	(0.53)	(0.14)	(0.15)	(0.2)
Constant	65.572	66.696	67.091	67.681	64.883	64.961	64.702
	(0.56)^***	(1.15)^***	(1.46)^***	(2.29)^***	(0.55)^***	(0.58)^***	(0.78)^***
R²	0.003	0.007	0.011	0.033	0.003	0.003	0.014
Obs	4,043	1,449	990	436	2,594	1,802	952
Panel B	Full Panel	Overdraft = 1			Overdraft = 0
		All	Age > 40	Age > 50	All	Age > 40	Age > 50
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
Wave	−0.093	−0.111	−0.183	−0.236	−0.086	−0.098	−0.125
	(0.05)^*	(0.09)	(0.11)^*	(0.15)	(0.06)	(0.07)	(0.09)
Health status	0.010	0.049	−0.021	0.093	−0.038	0.005	0.112
	(0.03)	(0.06)	(0.06)	(0.09)	(0.04)	(0.05)	(0.05)^**
Satisf. health	−0.100	−0.249	−0.238	−0.199	−0.008	−0.014	−0.026
	(0.14)	(0.25)	(0.21)	(0.28)	(0.16)	(0.18)	(0.2)
Prolonged illness	0.083	0.031	−0.051	−0.047	0.101	0.015	−0.281
	(0.13)	(0.18)	(0.18)	(0.29)	(0.18)	(0.22)	(0.32)
Constant	65.572	65.616	66.579	65.726	65.591	65.360	65.721
	(0.56)^***	(0.93)^***	(1.08)^***	(1.43)^***	(0.69)^***	(0.85)^***	(1.28)^***
R²	0.003	0.006	0.008	0.022	0.002	0.002	0.014
Obs.	4,043	1,356	924	411	2,687	1,868	977

Notes: The table shows the results of seven fixed-effects panel regressions with planned retirement age in years as the dependent variable. Data used for the analysis come from SAVE 2008, 2009, 2010, and 2011, encoded in the variable wave. All variables are as defined in the Appendix. All five imputations of the SAVE data are used. Coefficients and standard errors are calculated according to Rubin (1987). Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

Variable	Description
FAZ Survey
Planned retirement age	Age in years when nonretired participants plan to retire (“At what age do you plan to retire?”). Nonretirees only.
Actual retirement age	Age in years when retired participants did retire (“At what age did you retire?”). Retirees only.
Full retirement age	Full retirement age according to the German social security system based on the birth year of participants.
Retirement regret	Indicator that equals one if participants felt they retired too early (“If you were to retire again, would you retire earlier, later, or at the same age?”). Retirees only.
Inconsistency indicator	Refers to the time preference question as displayed in Figure 1. Indicator equals one if a participant makes at least one time-inconsistent choice in the direction of hyperbolic discounting.
Number inconsistent answers	Refers to the time preference question as displayed in Figure 1. Count of the number of inconsistent choices in the direction of hyperbolic discounting (0–3).
Impatience	Agreement on a seven-point Likert scale (1–7) with the statement “I am an impatient person.”
Age	Current age of participants in years.
Gender	Indicator that equals one if a participant is male.
Income	Self-reported monthly net income. Nonretirees only.
Retirement benefits	Retirement income from social and private insurance. Retirees only.
Satisfaction with benefits	Agreement on a seven-point Likert scale (1–7) with the statement “I am satisfied with my monthly retirement benefits.”
Number of children	Number of participant’s children.
Education	Variable that takes a value of two for a university degree or higher, a value of one for a high school degree (German Abitur), and a value of zero for a lower or no degree.
Married	Indicator that equals one if a participant is married.
Risk aversion	Agreement on a seven-point Likert scale (1–7) with the statement “I am a risk-averse person.”
Loss aversion	Agreement on a seven-point Likert scale (1–7) with the statement “I am very afraid of losses.”
Financial literacy	Number of correct responses to six financial literacy questions (see Online Appendix A).
Subjective life expectancy	Self-reported life expectancy in years of age.
Private pension insurance	Indicator if participant owns private pension insurance.
SAVE Survey
Planned retirement age	Age in years when nonretired participants plan to retire (“At what age do you plan to retire?”). Nonretirees only.
Actual retirement age	Age in years when retired participants did retire (“At what year did you enter retirement?”). Retirees only.
Smoker	Indicator that equals one if a participant responds yes to the question “Do you smoke regularly?” at least once in the used rounds of the panel.
Overdraft	Overdraft measures the self-reported usage of the overdraft option provided by the participant’s bank. It ranges from one (never use overdraft) to four (more than 6 times per year).
Overdraft indicator	Indicator that equals one if participants use the overdraft option frequently (overdraft ≥ 3) at least once in the used rounds of the panel.
Age	Current age of participants in years.
Gender	Indicator that equals one if a participant is male.
Income	Self-reported monthly net income.
Number of children	Number of participant’s children.
Education	Variable that takes a value of two for a university degree or higher, a value of one for a high school degree (German Abitur), and a value of zero for a lower or no degree.
Married	Indicator that equals one if a participant is married.
Financial literacy	Number of correct responses to nine financial literacy questions (see Online Appendix B).
Subjective life expectancy	Self-reported life expectancy in years of age.
Private pension insurance	Indicator that equals one if a participant owns private pension insurance.
Health status	Self-assessed health status on a five-point scale (from very good (=5) to very bad (=1)).
Satisfaction with health	Self-reported satisfaction with own health on a ten-point scale (from completely satisfied (=10) to not at all satisfied (=0)).
Prolonged illness	Indicator that takes the value of one if a participant reports to have prolonged health problems, illnesses, or disabilities.