econstor
Make Your Publications Visible.

A Service of

zbw
Leibniz-Informationszentrum
Wirtschaft
Leibniz Information Centre
for Economics

Friehe, Tim; Pannenberg, Markus

Working Paper

Time preferences and political regimes: Evidence
from reunified Germany

GLO Discussion Paper, No. 306

Provided in Cooperation with:
Global Labor Organization (GLO)

Suggested Citation: Friehe, Tim; Pannenberg, Markus (2019) : Time preferences and political
regimes: Evidence from reunified Germany, GLO Discussion Paper, No. 306, Global Labor
Organization (GLO), Maastricht

This Version is available at:
http://hdl.handle.net/10419/191521

Standard-Nutzungsbedingungen:

Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen
Zwecken und zum Privatgebrauch gespeichert und kopiert werden.

Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle
Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich
machen, vertreiben oder anderweitig nutzen.

Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen
(insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten,
gelten abweichend von diesen Nutzungsbedingungen die in der dort
genannten Lizenz gewährten Nutzungsrechte.

Terms of use:

Documents in EconStor may be saved and copied for your
personal and scholarly purposes.

You are not to copy documents for public or commercial
purposes, to exhibit the documents publicly, to make them
publicly available on the internet, or to distribute or otherwise
use the documents in public.

If the documents have been made available under an Open
Content Licence (especially Creative Commons Licences), you
may exercise further usage rights as specified in the indicated
licence.

www.econstor.eu



Time preferences and political regimes:
Evidence from reunified Germany

Tim Friehe∗ Markus Pannenberg†

January 21, 2019

Abstract

We use the separation and later reunification of Germany after World War II to show
that a political regime shapes time preferences of its residents. Using two identification
strategies, we find that former residents of the German Democratic Republic exhibit a
significantly less pronounced present bias when compared to former residents of the Fed-
eral Republic of Germany, whereas measures of patience are statistically indistinguishable.
Interpreting the years spent under the regime as a proxy for treatment intensity yields con-
sistent results. Moreover, we present evidence showing that present bias predicts choices
in the domains of health, finance, and education, thereby illustrating lasting repercussions
of a regime’s influence on time preferences.

Keywords: Time preferences; Political regime; Germany; Natural experiment; SOEP

JEL: D02, D12

∗University of Marburg, Public Economics Group, Am Plan 2, 35037 Marburg, Germany. CESifo, Munich,
Germany. EconomiX, Paris, France. E-mail: tim.friehe@uni-marburg.de.
†University of Applied Sciences Bielefeld, Department of Business and Economics, Interaktion 1, 33619

Bielefeld, Germany. IZA, Bonn, Germany. GLO, Maastricht, Netherlands. E-mail: markus.pannenberg@fh-
bielefeld.de.



1 Introduction

Time preferences are crucial to almost all major life choices such as saving for the future,

educational decisions, and both labor market and health-related behavior (e.g., DellaVigna

and Paserman 2005, Falk et al. forthcoming, Golsteyn et al. 2014, Koch et al. 2015, Meier

and Sprenger 2010, Sutter et al. 2013). At a national level, the time preferences of a country’s

residents are a key driving factor of domestic economic development (Dohmen et al. 2018).

As a result, a variation in time preferences will have very important implications at both the

individual and societal level.

This paper explores whether the socialist regime of the German Democratic Republic (GDR)

has caused a lasting variation in time preferences of its former residents. We use survey measures

from the German Socio-Economic Panel (SOEP) that are significantly correlated with time

preference information from incentivized experiments. To identify the time preference effect of

the GDR’s political regime, we use two identification strategies. First, following the standard

approach in the literature (e.g., Alesina and Fuchs-Schündeln 2007, Dragone and Ziebarth 2017),

we treat the reunification after more than four decades of separation as a natural experiment

and present findings from ordinary least squares regressions. To strengthen these results, we

test for robustness against bias from local unobservables applying the procedure by Oster

(forthcoming). Second, we adopt a geographic regression discontinuity design with border-

segment fixed effects, thereby addressing concerns about unobserved local heterogeneity by

comparing only people who have lived close to the former FRG/GDR border.

Traditionally, economists explain behavior for fixed preferences (e.g., Stigler and Becker

1977, Golsteyn and Schildberg-Hörisch 2017). Meier and Sprenger (2015) and Hardardottir

(2017) provide evidence in favor of the hypothesis that time preferences are rather stable over

the short- and medium-term.1 In contrast, we propose that the exposure to the GDR’s political

regime over a very long time horizon influenced its former residents’ time preferences in the

spirit of the literature arguing that preferences are endogenous to institutions (e.g., Bowles

1998, Fehr and Hoff 2011).

With respect to time preferences, we can, at an individual level, distinguish between present

bias and patience using our experimentally validated survey measures, via reference to the well-

established (β,δ) model as a parsimonious representation of time preferences that accounts

for the empirically important present bias (e.g., DellaVigna 2009, Laibson 1997, O’Donoghue

and Rabin 2015). In the (β,δ) model, the parameter β represents a bias for the present –

measuring how now is evaluated relative to later – whereas parameter δ is the discount factor

used in standard exponential discounting. Wang et al. (2016) identify information about time

1There is similar evidence, for example, regarding social preferences in Carlsson et al. (2014).

1



preferences in 53 countries, finding that present bias is a global phenomenon and much more

heterogeneous across countries than measures of patience. Similarly, Falk et al. (forthcoming)

describe notable variation in preferences within- and between-countries. By highlighting the

implications of past institutions on today’s time preferences, our analysis seeks to contribute

to an explanation of the origins of such heterogeneity in economic preferences.

According to our empirical results, the GDR’s socialist regime significantly diminished the

present bias of its former citizens in the long run. In contrast, the regime’s influence on patience

is insignificant.2 When we split the sample into birth cohorts, we find results that are consistent

with the proposition that a long exposure to the socialist regime during the years in which time

preferences are particularly malleable was necessary for a notable effect on present bias at the

individual level.

Against this background, we emphasize that the GDR’s socialist regime continues to shape

the lives of its former citizens to this day and will continue to do so into the future. The lasting

footprint is due, inter alia, to the intergenerational transmission of time preferences and social

norms (e.g., Brosig-Koch et al. 2011, Brown and van der Pol 2015, Chowdhury et al. 2018,

Gauly 2017). To assess the importance of our result of regime influence on present bias for

economic behavior, we show the predictive power present bias has for personal choices in the

health, finance, and education domains.

Our paper is the first to scrutinize the influence of political regimes on time preferences,

taking advantage of the German separation and reunification as a treatment variation. The

paper is thus related to contributions that search for past institutions’ footprints on behavior

and studies that similarly make use of the German separation and reunification as a natural

experiment.

There is recent literature on the long-term persistence and long-lasting effects of institu-

tions. For example, Becker et al. (2016) rely on the Life in Transition Survey to establish that

a historical affiliation with the Habsburg empire increases current levels of trust and reduces

corruption in courts and police in Central and Eastern Europe today. In other lines of inquiry,

Acemoglu et al. (2001) relate colonization styles to present economic performance, Nunn and

Wantchekon (2011) explain present levels of trust in Africa with references to the slave trade,

and Voigtländer and Voth (2012) find that pogroms in medieval times predict anti-Semitic vio-

lence in Nazi Germany. In comparison to these and related important contributions (see Bisin

and Verdier 2011 for a recent survey), our analysis considers a more recent set of institutions,

but still considers the possibility of formative institutional impact on individuals.

2The difference in the means of patience is smaller than the one for present bias, but in the anticipated
direction. The same holds for the signs of the (insignificant) GDR regression coefficient in the estimation for
the level of patience. Our data is thus consistent with the relatively greater variability of present bias across
countries as reported in Wang et al. (2016).

2



There are a number of recent contributions using the German separation and reunification to

form a better overall understanding of attitudes and economic behavior. For example, Brosig-

Koch et al. (2011) present experimental findings indicating that East Germans exhibit less

solidarity at the private level than West Germans (thereby revisiting the research question of

Ockenfels and Weimann 1999). Consistent with the idea that East Germans perceive the state

to be the responsible actor in that domain, Alesina and Fuchs-Schündeln (2007) show that

East Germans support redistribution and state intervention more than West Germans. These

authors also highlight that the difference will remain for a considerable period of time. These

lasting consequences may be attributed to social norms and intergenerational transmission.

Necker and Voskort (2014) study intergenerational transmission of attitudes, focusing on East

and West Germany and the responses of parents and children to value questions (e.g., whether

owning a house is important). Bauernschuster and Rainer (2012), Beblo and Görges (2018), and

Campa and Serafinelli (forthcoming) are interested in the long-run impact of regime differences

for labor market participation, and Friehe and Mechtel (2014) in potential implications for the

relevance of conspicuous consumption. In contrast, the present paper seeks to identify regime

repercussions on economic preferences that, in turn, influence many kinds of behavior.

Our paper is closely related to Rainer and Siedler (2009) and Heineck and Süssmuth (2013)

who examine differences between East and West Germans regarding the level of trust, which

can be interpreted as a measure of social preferences (e.g., Becker et al. 2012). Like our study,

their contributions rely on the recent German history for identification of causal effects and

the SOEP as one data source.3 Rainer and Siedler (2009) and Heineck and Süssmuth (2013)

find that East Germans show higher levels of social distrust. The latter paper additionally

considers risk preferences of former GDR and FRG citizens, and concludes that East Germans

are similarly risk averse (after a brief period of lower risk aversion).

Economic preferences and personality traits are complementary when it comes to explaining

heterogeneity in life outcomes and behavior (e.g., Humphries and Kosse 2017). Consistent with

this, when considering correlations between six personality measures and time preferences in a

large SOEP data set, Becker et al. (2012) find only two significant instances: one in which the

correlation with conscientiousness is only of small magnitude, and another with agreeableness

correlated with a medium magnitude. However, there exist conceptual relationships. For exam-

ple, Almlund et al. (2011: p. 70) refer to a conceptual relationship between conscientiousness

and time preferences. In our robustness checks, we thus seek to test whether the GDR treat-

ment effect also shows up in selected personality traits which are related to time preferences.

For this purpose, we build on Friehe et al. (2015) by providing refined ordinary least squares

and new geographic discontinuity design results for both conscientiousness and agreeableness

3Rainer and Siedler (2009) primarily build on the German General Social Survey (ALLBUS).

3



in one of our robustness checks. Friehe et al. (2015) analyze how the GDR regime influenced

personality traits within an ordinary least squares framework.

The structure of the paper is as follows. Section 2 presents the data. In Section 3, we

will explain our research design. Section 4 presents our empirical analysis including robust-

ness checks. Section 5 explores some important implications from our main result. Section 6

concludes.

2 Data

Our analysis is based on the German Socio-Economic Panel (SOEP), a nationally representative

longitudinal data set created in 1984. Our working sample contains respondents who were either

born in Germany or immigrated before 1949 and have provided valid information about where

they lived in 1989 (i.e., the year before reunification). These data selection criteria follow

from the treatment of the separation and reunification of East and West Germany as a natural

experiment.

Specifically, we use the information on patience and impulsivity from the SOEP waves 2008

and 2013. The question regarding patience asks: “Are you generally an impatient person, or

someone who always shows great patience?”. Answers are coded on an 11-point scale, with 0

denoting “very impatient” and 10 “very patient”. The question concerning impulsivity asks:

“Do you generally think things over for a long time before acting - in other words, are you

not impulsive at all? Or do you generally act without thinking things over for a long time - in

other words, are you very impulsive?”. Answers are again coded on an 11-point scale, with 0

denoting “not at all impulsive” and 10 “very impulsive”.

Our theoretical reference with respect to time preferences is the (β,δ) model. In that

framework, intertemporal preferences from the perspective of period t can be represented by

Ut = ut +
∑T

τ=t+1 βδ
τuτ with δ as the standard discount factor, such that β = 1 corresponds to

exponential discounting and while β ∈ (0, 1) reflects present bias (e.g., O’Donoghue and Rabin
2015).4 Answers to the question regarding patience represent our proxy of the δ component of

the (β,δ) model. The patience information was validated by Vischer et al. (2013) by reference

to experimentally elicited information on time preferences. Answers to the question regarding

impulsivity represent our proxy of the β component of the (β,δ) model. This interpretation

is sensible for three reasons. First, our chosen SOEP impulsivity question matches items in

the Barratt impulsiveness scale (Patton et al. 1995) and the Tangney et al. (2004) self-control

scale. The terms “present bias” and “self-control issues” are often used synonymously (see,

4Present-biased preferences thus give special importance to immediate payoffs and allow for time-inconsistent
choice (e.g., O’Donoghue and Rabin 1999).

4



e.g., DellaVigna 2009). Indeed, since greater self-control leads to a greater congruence between

actions an agent would like to undertake and those actually undertaken, it may be interpreted

as a less pronounced present bias (e.g., Ameriks et al. 2007). Second, answers to the impulsivity

question correlate with experimentally elicited present bias (see Appendix A for details of our

analysis).5 Along the same lines, Burks et al. (2012) demonstrate a significant correlation

between experimental measures of impulsivity and present bias conditional on experimental

information on patience and a survey measure of impatience. Third, present bias information

from answers to the impulsivity question correlates – as it should – with behaviors critically

influenced by self-control issues such as smoking (see Section 5). Accordingly, we use the SOEP

impulsivity measure as a proxy of present bias in our empirical work. Note, as example, that

Fourage et al. (2014) similarly refer to the patience and impulsivity questions in terms of the

(β,δ) model.

The covariate of key interest in our empirical analysis is a dummy variable that is equal

to one (zero) if the respondent was a resident of the GDR (FRG) in 1989. We include a

host of further covariates that may be relevant to the formation of time preferences but are

not themselves influenced by the individual’s time preferences (see Table 1). The age of the

respondent is included (Bishai 2004). Gender is included as a dummy variable equal to one when

the respondent is male and zero otherwise, as it has been shown to bear on time preferences

(Bishai 2004). Because preferences are very much shaped by one’s parents and environment

during childhood (e.g., Kosse et al. 2016), we also include information about whether or not the

respondent was raised in a small, medium-sized, or large community and about the educational

background of the parents (for which we use dummy variables for the highest degrees obtained

by the father and the mother).6

Table 1 presents summary statistics by treatment status for the (non-standardized) variables

used in our empirical analysis. The descriptive results indicate that former GDR and FRG

residents differ with respect to some covariates, suggesting adjustments for covariate differences

in our empirical specifications. Imbens and Wooldridge (2009) point out that differences in the

observable characteristics of the treatment group and the control group might lead to sensitive

estimation results in a linear regression framework. They propose that the imbalance of the

covariate distributions should be assessed by testing whether or not the scale-free normalized

difference between treatment and control group covariate means exceeds 0.25 (as a rule of

thumb). The normalized differences in our data are less than 0.25 for all but one covariate in

5Pinger (2017) also finds that survey measures and experimental measures of present bias are significantly
related. She uses the special experimental SOEP submodul 2006 and a survey measure of present bias that is
not available in the main SOEP waves.

6Including the educational background of the parents is also important because own cognitive ability relates
to time preferences, and is itself strongly influenced by parents (Anger and Heineck 2010, Anger and Schnitzlein
2017, Bishai 2004, Dohmen et al. 2010).

5



Column (3) of Table 1 (and in this single instance, Mother: No vocational degree, the difference

is very close to 0.25).

(1) (2) (3)

FRG GDR Absolute Value

Count Mean Count Mean Normalized Difference

Age 27,784 51.11 11,208 49.95 0.050

Male 27,784 0.467 11,208 0.457 0.015

Raised in a large city 27,784 0.232 11,208 0.195 0.063

Raised in a medium city 27,784 0.169 11,208 0.180 0.019

Raised in a small city 27,784 0.211 11,208 0.242 0.052

Mother: Secondary lower school 27,784 0.648 11,208 0.542 0.154

Mother: Secondary intermediate school 27,784 0.171 11,208 0.273 0.175

Mother: Secondary higher school 27,784 0.0656 11,208 0.0750 0.026

Mother: No vocational degree 27,784 0.365 11,208 0.198 0.267

Mother: Vocational degree 27,784 0.475 11,208 0.561 0.123

Mother: Technical school 27,784 0.00576 11,208 0.0469 0.183

Mother: University degree 27,784 0.0381 11,208 0.0622 0.078

Father: Secondary lower school 27,784 0.611 11,208 0.519 0.131

Father: Secondary intermediate school 27,784 0.128 11,208 0.224 0.179

Father: Secondary higher school 27,784 0.132 11,208 0.111 0.045

Father: No vocational degree 27,784 0.110 11,208 0.0562 0.139

Father: Vocational degree 27,784 0.675 11,208 0.652 0.035

Father: Technical school 27,784 0.0127 11,208 0.0412 0.125

Father: University degree 27,784 0.105 11,208 0.114 0.019

Notes: We use SOEP data from 2008 and 2013. Our working sample contains respondents who were

either born in Germany or immigrated before 1949 and have provided valid information about where

they lived in 1989 (i.e., the year before reunification). For each covariate, the normalized difference

is defined as the difference in averages by treatment status, scaled by the square root of the sum of

variances.

Table 1: Descriptive Statistics

3 Research Design

We propose that any differences in the time preferences of former GDR and FRG residents

measured after the country’s reunification are related to how the two different political regimes

treated their citizens during the 40-plus years of separation. We pursue two identification

strategies to present robust evidence for the hypothesized causality that we explain in some

detail in Section 3.2 after we have discussed the malleability of time preferences across a lifespan

and explain hypotheses concerning the influence of the regime in this regard in Section 3.1.

6



3.1 The Formation of Time Preferences and Potential Regime In-

fluence

Time preferences as well as other individual attributes are heavily influenced by parents through

genetics, parental investment, and the choice of childhood environments (e.g., Anger and Schnit-

zlein 2017, Cunha and Heckman 2007, Dohmen et al. 2012). According to the literature on

cultural transmission, parents invest into the formation of the values and preferences of their

children’ in order to maximize welfare (e.g., Bisin and Verdier 2011, Bulte and Horan 2011,

Tabellini 2008).7 Doepke and Zilibotti (2008), for example, focus on parental desire to instill

patience. Parents may – at a private cost – choose to instill preferences different from their

own when these preferences serve their children better. This signifies that parents raised in a

past political regime may choose to induce values and preferences optimally adjusted to the

current political regime. In a dynamic context, parents may have different expectations about

the evolution of the regime, which would contribute to preference heterogeneity among chil-

dren. With respect to the intergenerational transmission of time preferences, Chowdhury et

al. (2018) find evidence in a developing-country context and Gauly (2017), Kosse and Pfeiffer

(2013), and Brenoe and Epper (2018) do so for developed countries. The latter highlight that

there is evidence that this transmission is mainly due to parental investment instead of genetics.

This supports the cultural transmission approach, which also very much emphasizes that the

interaction of individuals with teachers, friends, peers, and role models critically influence the

formation of preferences (e.g., Behabib and Bisin 2011, Bisin and Verdier 2011).8

The malleability of cognitive and non-cognitive skills is particularly high in early childhood

years (e.g., Fehr et al. 2008). For example, Knudsen et al. (2006) refer to evidence from different

disciplines including economics and neurobiology. Cunha and Heckman (2007) highlight that

most measurable skill gaps open up at early ages and show near parallelism during school years.

Deckers et al. (2017) similarly maintain that differences in cognitive and non-cognitive abilities

emerge early in life, alluding to critical periods for preference formation.9 With respect to time

preferences, we can refer to evidence presented in Sutter et al. (2013, 2015) and Bettinger

and Slonim (2007). Sutter et al. (2013) study how experimentally-elicited time preferences

are related to real-life outcomes for a sample of individuals of ages 10 to 18. They find that

time preferences are already relatively stable during this time period. In contrast, Sutter et al.

7Zumbuehl et al. (2013) and Dohmen et al. (2012), for example, provide strong empirical support for the
cultural transmission approach.

8Evolutionary selection also plays a role in the formation of preferences but is less relevant for our context.
The literature is surveyed by Robson and Samuelson (2011), for example.

9Delaney and Doyle (2012) present evidence on the relationship between psychometric measures proxying
time preferences and socioeconomic differences that is consistent with the findings reported in Deckers et al.
(2017).

7



(2015) analyze choices of children between the ages of 3 and 6 and find that time preferences are

malleable during this period. Consistent with this, Bettinger and Slonim (2007) report that,

whereas younger children have higher discount rates than adults, individual discount rates

by the age of 16 years are fairly similar to those of adults. However, as holds true for other

attributes, time preferences may be subject to change at later points as well. The impressionable

years hypothesis proposes that core attitudes, beliefs, and values crystallize during a period of

great mental plasticity in early adulthood and remain largely unaltered thereafter (Krosnick

and Alwin 1989). Giuliano and Spilimbergo (2014) overlay the impressionable years hypothesis

to life experience data, focusing on the range 18 to 25 years of age and the repercussions from

experiencing a recession during those years. With regard to later changes (i.e., those in middle

age), it is usually argued that they lack economic significance (e.g., Cobb-Clark and Schurer

2013). In contrast, changes at old age may be important again (e.g., Specht et al. 2014).

Using our data, we can explore the extent to which our analyzed time preference measures

change and whether this differs across age cohorts. In general, in exploring the differences

between each subject’s responses in 2008 and 2013 for the 65% of the sample for whom we have

observations in both years, we find that the median differences are zero, that 23% (54%) of

respondents have a difference of zero (of an absolute level less than or equal to one) with regard

to impulsivity, and that 25% (57%) of respondents have a difference of zero (of an absolute

level less than or equal to one) with respect to patience. To further describe the stability, we

follow the approach of Cobb-Clark and Schurer (2013) and consider how the difference between

the measurement in 2013 and the one in 2008 at the level of the individual is a function of the

individual’s age (see Figure 1).10 Overall, our data is very much consistent with the argumenta-

tion presented above. There are meaningful changes in early adulthood – as emphasized by the

impressionable years hypothesis, for example – and there tends to be some variability in old age

again. However, in middle age, stability is very high, particularly for our present bias measure.

When it comes to intra-individual changes as portrayed in Figure 1, Schildberg-Hörisch (2018),

for example, argues that it is meaningful to think about economic preferences as a latent trait

that is measurable only with some noise induced by contemporaneous emotions.

10Non-parametric local mean smoothing is applied to estimate the changes by age.

8



-.5-.250.25.5
change of impulsivity

2
0

2
5

3
0

3
5

4
0

4
5

5
0

5
5

6
0

6
5

7
0

7
5

8
0

8
5

9
0

a
g
e

N
o

te
: 

S
O

E
P

 2
0

0
8

/2
0

1
3

- 
fu

ll 
sa

m
p

le
 -

A
g
e
 p

ro
fil

e
 o

f 
in

d
iv

id
u
a
l 5

-y
e
a
r-

ch
a
n
g
e
s 

o
f 
p
re

se
n
t 
b
ia

s

-.5-.250.25.5
change of impulsivity

2
0

2
5

3
0

3
5

4
0

4
5

5
0

5
5

6
0

6
5

7
0

7
5

8
0

8
5

9
0

a
g
e

N
o

te
: 

S
O

E
P

 2
0

0
8

/2
0

1
3

- 
co

n
tr

o
l g

ro
u

p
 -

A
g
e
 p

ro
fil

e
 o

f 
in

d
iv

id
u
a
l 5

-y
e
a
r-

ch
a
n
g
e
s 

o
f 
p
re

se
n
t 
b
ia

s

-.5-.250.25.5
change of impulsivity

2
0

2
5

3
0

3
5

4
0

4
5

5
0

5
5

6
0

6
5

7
0

7
5

8
0

8
5

9
0

a
g
e

N
o

te
: 

S
O

E
P

 2
0

0
8

/2
0

1
3

- 
tr

e
a

tm
e

n
t 

g
ro

u
p

 -
A

g
e
 p

ro
fil

e
 o

f 
in

d
iv

id
u
a
l 5

-y
e
a
r-

ch
a
n
g
e
s 

o
f 
p
re

se
n
t 
b
ia

s

-.5-.250.25.5
change of patience

2
0

2
5

3
0

3
5

4
0

4
5

5
0

5
5

6
0

6
5

7
0

7
5

8
0

8
5

9
0

a
g
e

N
o

te
: 

S
O

E
P

 2
0

0
8

/2
0

1
3

- 
fu

ll 
sa

m
p

le
 -

A
g
e
 p

ro
fil

e
 o

f 
in

d
iv

id
u
a
l 5

-y
e
a
r-

ch
a
n
g
e
s 

o
f 
p
a
tie

n
ce

-.5-.250.25.5
change of patience

2
0

2
5

3
0

3
5

4
0

4
5

5
0

5
5

6
0

6
5

7
0

7
5

8
0

8
5

9
0

a
g
e

N
o

te
: 

S
O

E
P

 2
0

0
8

/2
0

1
3

- 
co

n
tr

o
l g

ro
u

p
 -

A
g
e
 p

ro
fil

e
 o

f 
in

d
iv

id
u
a
l 5

-y
e
a
r-

ch
a
n
g
e
s 

o
f 
p
a
tie

n
ce

-.5-.250.25.5
change of patience

2
0

2
5

3
0

3
5

4
0

4
5

5
0

5
5

6
0

6
5

7
0

7
5

8
0

8
5

9
0

a
g
e

N
o

te
: 

S
O

E
P

 2
0

0
8

/2
0

1
3

- 
tr

e
a

tm
e

n
t 

g
ro

u
p

 -
A

g
e
 p

ro
fil

e
 o

f 
in

d
iv

id
u
a
l 5

-y
e
a
r-

ch
a
n
g
e
s 

o
f 
p
a
tie

n
ce

F
ig

u
re

1
:

D
iff

er
en

ce
s

b
et

w
ee

n
T

im
e

P
re

fe
re

n
ce

M
ea

su
re

2
0
1
3

a
n
d

2
0
0
8

A
cr

o
ss

C
o
h
o
rt

s
(9

5
%

C
o
n
fi
d
en

ce
In

te
rv

a
ls

A
re

D
is

p
la

y
ed

).

9



Economists have traditionally used exogenous preferences to explain how behavior changes

when constraints change. Focusing on patience, Becker and Mulligan (1997) present a model

in which time-consistent individuals can invest into their own time preferences to overcome

impatience as a potential human fallibility, where the investment involves time and effort spent

on active appreciation of future pleasures.11 A high investment in patience is more likely the

lower current marginal utility of wealth (as investment implies giving up current consumption)

and the higher future utility of future wealth (as greater patience makes future utilities more

salient in the present). Their setup uses the exponential discounting model such that introduc-

ing a potential present bias that can also be influenced by investment moves it closer to our

setup. Endogenous present bias as an additional channel for the decision-maker is very real-

istic due to evidence that people can improve their self-control by investment in conscientious

behavior (e.g., Baumeister and Tierney 2012, Mischel 2014, Muraven 2012). Importantly, it

adds a strategic rationale for investment to the discussion by Becker and Mulligan (1997): An

individual sophisticated about the present bias may invest to improve upon future decisions

made at a later point in life.

Against this background, we now conjecture a possible regime influence on time preferences.

With the regime’s astute consideration of the early life development of preferences and values,

political education in the GDR started in day-care centers and was very important in school.

The ultimate goal of the GDR’s political education was creating socialist citizens characterized

by a high sense of duty and self-discipline (e.g. Litz 2007). The regime idealized a forward-

looking perspective. For example, Walter Ulbricht announced ten commandments during the

fifth SED party congress in June 1958 including the admonishment that people should live

sparingly and strive for improvement of their performance in the future. This indoctrination

thus inculcates greater patience and a less pronounced present bias of former GDR residents.

The GDR’s massive state-security service, which relied extensively on both a network of un-

official collaborators and extreme forms of repression, strongly incentivized GDR residents to

control their impulses. Indeed, people were on constant alert regarding statements they could

make and actions they could undertake without endangering their personal freedom or their

physical wellbeing (e.g., Bruce 2010, Fulbrook 2005). The cultural transmission model would

thus predict that parents go to extreme lengths to instill self-control in their children because

it was of overriding importance for life in the GDR. This accords with the incentives to invest

in own preferences as analyzed and presented by Becker and Mulligan (1997). With respect to

patience, GDR citizens had to cope with shortages in a myriad of life domains including hous-

ing, vacations, and consumption goods. The model presented by Becker and Mulligan (1997)

11In this context, it is interesting that Chen (2013) documents that people who speak languages that do not
delineate the future tense, and thus allowing “tomorrow” to be a continuation of “today” instead of forcing it
to be a new day altogether – are more patient.

10



explicitly accounts for the possibility that GDR residents put forth more effort to increase

their appreciation of the future in order to optimally respond to the shortages and delivery

times. The cultural transmission model would similarly inform a higher relative expectation of

patience among former GDR residents as a result of parental investment decisions. On a dif-

ferent note, when comparing the political regimes in the GDR and the FRG, it unambiguously

results that the GDR ensured more stable, less heterogeneous economic positions of individu-

als. This resulted because, among other factors, the GDR featured no official unemployment

and a very compressed wage distribution.12 Sozou (1998) and Dasgupta and Maskin (2005)

provide theoretical setups showing that uncertainty about either the occurrence or the precise

timing of payoffs in the future can produce preferences consistent with present bias. The rel-

ative absence of notable uncertainty regarding the individual economic position in the GDR

thus suggests that the GDR’s political regime may have lowered the present bias of its former

residents relative to that of FRG residents.

In summary, there is evidence that points to malleability of preferences particularly in

early life. Moreover, we find that different approaches lead to the hypothesis that former

GDR residents are characterized by a less pronounced present bias and greater patience when

compared to former FRG residents. Next, we address how we test these hypotheses with the

data available.

3.2 Identification Strategy: The German Separation and Reunifica-

tion as a Natural Experiment

Our identification strategies rely on treating the German separation into the FRG and the

GDR as a natural experiment. Fuchs-Schündeln and Hassan (2016) forcefully argue that the

separation of Germany was exogenous to the preferences of the affected populations and the

economic conditions in East and West Germany at the time. Ideally, the following conditions

are fulfilled: (i) the treatment was exogenous, (ii) East and West Germans (i.e., the treated and

the non-treated) were comparable when the treatment was imposed, (iii) there was no selective

migration throughout the treatment period, and (iv) there was trackable migration after the

reunification. We will now address these issues in turn.

Imposition of the Treatment The imposition of the political regimes was truly exogenous.

The geographical division of Germany was a result of the agreement between the United King-

dom, the United States, and the Soviet Union on the partition of postwar Germany in 1944,

12The political regime of the GDR vigorously highlighted differences between the FRG and the GDR with
respect to unemployment, for example. The salience of such differences for its residents was aided by the use of
propaganda in school books or shows on television (e.g., Saunders 2007).

11



which split the territory into three sectors of roughly equal population size.13 Redding and

Sturm (2008) highlight that the decisions determining the partition of East and West Germany

are unlikely to be correlated with the prewar characteristics of respective regions. Moreover,

Wolf (2009) states that by the end of the Weimar Republic in 1933, Germany was an econom-

ically well-integrated area, such that the separation into East and West Germany that existed

between about 1946 and 1989 could not have been predicted in 1939.

Comparability of East Germans and West Germans Before the Treatment A key

identifying assumption is that former GDR and FRG residents did not differ from each other

with regard to time preferences prior to the German separation. While there is no representa-

tive data on this issue, there is documented similarity between East and West Germans with

respect to a host of other variables prior to the separation. Alesina and Fuchs-Schündeln (2007)

explain that the regions that became the FRG and the GDR were similar in terms of pre-World

War II average per-capita income levels and the amount of destruction experienced during

World War II. East and West Germany were also similar in terms of the proportions of the

working population involved in various industries (Schäfgen 1998), the political orientations of

voters at the turn of the century (Alesina and Fuchs-Schündeln 2007), and population densi-

ties (Hubert 1998). Comparability with respect to per-capita income levels and the amount of

war-related destruction is important for our design, for example, as there is some evidence that

the availability of financial resources impacts intertemporal decision-making (e.g., Carvalho et

al. 2016). Similarly, the split of the working population may be relevant to time preferences

as Nguyen (2011) finds some relationship between time preferences and occupations. Based on

such evidence, Fuchs-Schündeln and Hassan (2016), among others, conclude that the two parts

of Germany were indistinguishable prior to the separation.

Selective Migration During the Separation Period After the two political regimes had

been imposed, around three million people emigrated from the GDR to the FRG before the

Berlin Wall was built in August 1961 (e.g., Heidemeyer 1994, Hubert 1998). There was very

little migration from the GDR to the FRG after 1961 or from West to East Germany. Intellec-

tuals and entrepreneurs were overrepresented among the sample of migrants (e.g., Heidemeyer

1994).14 Migration during our treatment period may confound our interpretation of results

13Later on, the US and the UK handed over small parts of their sectors to France (e.g., Burchardi and Hassan
2013).

14However, Schäfgen (1998: 58), for instance, asserts that the structure of society with respect to educational
achievement and implied differences between social groups remained relatively stable and comparable in East
and West Germany.

12



when migrants’ time preferences were systematically different from those of non-migrants.15

When implementing the first identification strategy, we include information about an individ-

ual’s parents, in an attempt to address potential selection effects arising from migration after

the imposition of the GDR’s socialist regime but before the erection of the Berlin Wall. For a

subsample of the data, we can actually account for individuals who migrated to West Germany

during the treatment period and find results consistent with our main results.

Migration After the Reunification After the fall of the GDR’s socialist regime, we have

full data with regard to migration between East and West Germany.16 In particular, our data

set allows us to differentiate between East Germans who lived in the GDR in 1989 and continue

to live in that region of Germany today from those who lived in the GDR in 1989 but moved

to the west before their participation in the SOEP survey. Similarly, we can track movements

from West Germany to East Germany.

In summary, we believe that the identifying assumption that differences in time preferences

were shaped by the respective experiences of populations under the two political regimes (par-

alleling the approach taken by Alesina and Fuchs-Schündeln (2007) and others) seems justified.

3.2.1 Ordinary Least Squares Regressions and Test of Selection on Unobservables

Based on our discussion above, we simply compare former GDR with former FRG residents

using ordinary least squares regressions with (arguably) exogenous covariates as our first iden-

tification strategy. The covariates used in our regression exercises are summarized in Table

1. Note that including information pertaining to the survey respondent’s parents allows us to

control for the characteristics of the individuals responsible for a potential migration decision

from East to West during the years of the German separation. Nevertheless, one might still be

concerned that our results may be prone to bias from local unobservables, for example, since

we do not have information on the distribution of time preferences before the separation of

Germany. Therefore, we bolster our results from ordinary least squares regressions by testing

against omitted variables bias using the procedure suggested by Oster (forthcoming).

3.2.2 Geographic Regression Discontinuity Design

In our setting, the GDR treatment changes discontinuously at the former FRG/GDR border.

The basic idea of using a geographic regression discontinuity design (GRDD) is that individuals

15Migration choices may be related to time preferences. For example, Goldbach and Schlüter (2018) present
evidence that migration from developing countries is more likely for patient individuals.

16See, for example, Hunt (2006) and Fuchs-Schündeln and Schündeln (2009) for a description and analysis of
migration after the end of the GDR.

13



who lived in former FRG counties close to the border represent good comparisons for individuals

who lived in GDR counties close to the border, since spatially close counties were in all likelihood

more similar to each other before the political division. Accordingly, the procedure considers

only observations within a specific bandwidth around the border and – with respect to the

observations within the bandwidth – attaches more weight to individuals living closer to the

FRG/GDR border. We will consider three different bandwidths: The mean squared error

optimal bandwidth suggested by Imbens and Kalyanaraman (2012) as well as two multiples of

this optimal bandwidth (0.75, 1.25).

The running variable in our empirical model is the Euclidian distance from the centroid

of each respondent’s county of residence and the border.17 Following Gelman and Imbens

(forthcoming), for example, we estimate the following local linear regression with a triangular

kernel

yicbt = α + βGDRi + γ1Euclidc + γ2Euclidc ∗GDRi + δb + �icbt, (1)

where the dependent variable yicbt is the time preference measure of individual i living in county

c along the border segment b at time t. GDRi is an indicator for having lived in the GDR

in 1989 (i.e., before reunification). Euclidc is the Euclidian distance as defined above and

the interaction Euclidc ∗ GDRi allows for different slopes of the regression function on both
sides of the FRG/GDR border. By including δb, we incorporate fixed effects for eight equally-

sized border segments to allow for heterogeneous levels of the outcome of interest along the

FRG/GDR border (as in Dell 2010).18 Finally, �icbt is an error term that can be considered as

a random error generating variation in yicb around the local regression line. We compute both

standard errors clustered at the individual level (due to the unbalanced nature of our working

sample) and standard errors clustered at the county level (to take into account the discreteness

of our running variable; see, e.g., Lee and Card 2008).

Our GRDD is similar to those presented in the literature, but we want to stress two impor-

tant differences.19 First, our outcome measures stem from the years 2008 and 2013, allowing

for observed sorting around the former border after reunification but before our years of ob-

servation. To account for this concern, we drop survey respondents who moved from West to

East or from East to West Germany within the relevant post-reunification period in our GRDD

specifications.20 Second, the potential role of specific government policies during the period of

17We incorporate 396 different counties in the estimating sample for our GRDD analysis.
18We exclude Berlin from our working sample for two reasons. First, we cannot calculate a reasonable

Euclidian distance for people living in West Berlin (as there was both a border surrounding Berlin and the
Berlin wall border). Second, we suspect strong selective migration in both parts of Berlin during the German
separation for various reasons (e.g., political and cultural ones).

19For example, Becker et al. (2016) also assess the long-run effects of a border that existed in the past.
20We thereby lose about 4.4 % of our estimating sample (of which 3.6 % represent East-West movers). The

results from our ordinary least squares regression exercises that we will report in Section 4.1 are robust to this

14



the German separation must be acknowledged. Specifically, West Germany started a sizable

subsidization scheme in 1971 to counteract potential disadvantages from being cut-off from

adjacent markets due to the FRG/GDR border and to curtail possible emigration. The FRG

thereby created the Zonenrandgebiet including all districts that have 50 percent of either their

area or their population within a distance of 40 km to the border to the GDR or Czechoslo-

vakia (see, e.g., von Ehrlich and Seidel forthcoming). In the GDR, access to areas close to

the border was always very restricted and there is evidence that people who were considered

politically unreliable (and others) were forced to move further away from the border. Moreover,

the area that was in a bandwidth of about 5 km was categorized as prohibited area implying

very strict rules with regard to such aspects as cultural life and visitation rights, on the one

hand, and privileges in terms of food stamps etc. on the other (e.g., BStU 2012). As a result

of these place-based policies, selective sorting along the border during the separation period

may be expected on both sides. To account for this concern, we also employ donut specifica-

tions in our empirical work. Specifically, we first present results for the optimal bandwidths

h∗ (0.75/1.25 ×h∗) according to the criterion suggested by Imbens and Kalyanaraman (2012).
In addition, we show parameter estimates from donut specifications with optimal bandwidths

h∗ (1.25 × h∗), where we do not consider observations from Western (East) counties with a
centroid less than 40 (10) km from the FRG/GDR border.21

4 Empirical Results

Our findings on the influence of the GDR’s socialist regime on time preferences are presented

in this section. After presenting descriptive evidence on the matter, we will turn to results from

our two identification strategies and then discuss findings from robustness checks.

In Table 2, we see that the treatment group’s mean of the proxy for present bias is notably

lower than that of the control group, whereas the treatment group’s mean of the proxy for

patience does not differ significantly. Moreover, when we use indicator variables, which are

equal to one when the response on the respective preference measure was higher than six, we

find similar patterns. In both cases, t-test results indicate that present-bias is significantly less

pronounced in the treatment group.

change in the sample. Results are available upon request.
21In their study about work attitudes of females, Campa and Serafinelli (forthcoming) also apply a GRDD

approach to the natural experiment represented by the German separation/reunification episode, similarly
including donut specifications.

15



Present Bias Patience

Answer

Control group (FRG) 5.12 6.10

Treatment group (GDR) 4.95 6.13

Dummy Variable= 1 when Answer > 6

Control group (FRG) 0.29 0.50

Treatment group (GDR) 0.25 0.49

Notes: We use SOEP data from 2008 and 2013. Dummy variable is equal to

one when the response on the respective time preference measure was higher

than six. The numbers of observations are as follows: Present bias & FRG:

N = 23, 169, present bias & GDR: N = 8, 701, patience & FRG: N = 23, 184,

patience & GDR: N = 8, 695.

Table 2: Means for SOEP Proxies of Present Bias and Patience

4.1 Results from Ordinary Least Squares Regressions

Using ordinary least squares regressions, we find that former GDR citizens exhibit significantly

lower levels of present bias (Column (1) in Table 3). The sign of the treatment effect is consistent

with the idea that institutional aspects of the GDR (such as the induced economic stability

and the impulse-control incentivized by the massive state-security service) diminished present

bias. The size of the GDR treatment effect is comparable to that of the gender effect, and

the difference between the two is not significantly different from zero. In contrast, we do not

find a lasting influence of the GDR’s political regime on patience. The estimated parameter is

positive, as expected, but not significant.

16



(1) (2)

Present Bias Patience

GDR -0.197∗∗ 0.005

(0.030) (0.032)

Age 0.014∗∗ 0.022∗∗

(0.005) (0.005)

Age2 -0.000∗∗ -0.000∗∗

(0.000) (0.000)

Male -0.272∗∗ -0.009

(0.027) (0.029)

Raised in a large city 0.217∗∗ -0.103∗

(0.037) (0.040)

Raised in a medium-sized city 0.161∗∗ -0.030

(0.039) (0.042)

Raised in a small city -0.002 0.045

(0.036) (0.039)

Mother: Secondary lower school 0.088 -0.120+

(0.062) (0.066)

Mother: Secondary intermediate school 0.119+ -0.155∗

(0.068) (0.074)

Mother: Secondary higher school -0.023 -0.123

(0.094) (0.102)

Mother: No vocational degree -0.069 0.060

(0.053) (0.056)

Mother: Vocational degree 0.040 0.010

(0.052) (0.056)

Mother: Technical school -0.073 0.056

(0.105) (0.115)

Mother: University degree -0.112 0.021

(0.103) (0.112)

Father: Secondary lower school -0.144∗ 0.110+

(0.060) (0.065)

Father: Secondary intermediate school -0.173∗ 0.095

(0.068) (0.074)

Father: Secondary higher school -0.170∗ 0.098

(0.082) (0.090)

Father: No vocational degree 0.095 0.097

(0.070) (0.075)

Father: Vocational degree 0.105+ 0.070

(0.055) (0.060)

Father: Technical school 0.084 -0.091

(0.102) (0.114)

Father: University degree 0.138 0.065

(0.085) (0.091)

N 33,109 33,119

R2 0.016 0.007

Notes: We use SOEP data from 2008 and 2013. Parameter estimates come from ordinary least

squares regressions. The dependent variable is the SOEP measure for present bias and patience,

respectively. Robust standard errors are clustered at the individual level and presented in paren-

theses; + p < 0.10, ∗ p < 0.05, ∗∗ p < 0.01.

Table 3: Time Preferences and GDR: Ordinary Least Squares Regressions

Matching the results of earlier studies using the natural experiment of the German separa-

17



tion and reunification on the dynamics of the East-West divide (e.g., Brosig-Koch et al. 2011,

Rainer and Siedler 2009), we find that there is no notable convergence with regard to present

bias when considering the five-year lag between 2008 and 2013. More specifically, including an

interaction term of GDR and the year 2013 in our specifications from Table 3, we find that

there is no remarkable change in the estimated parameters of interest and that the coefficient

of the interaction term is never significantly different from zero (see Table 4).

(1) (2)

Present Bias Patience

GDR -0.223∗∗ -0.028

(0.036) (0.039)

Year 2013 0.092∗∗ 0.063∗

(0.025) (0.026)

GDR × 2013 0.056 0.071
(0.044) (0.045)

N 33,109 33,119

R2 0.016 0.007

Notes: We use SOEP data from 2008 and 2013. Pa-

rameter estimates come from ordinary least squares

regressions. The dependent variable is the SOEP

measure for present bias and patience, respectively.

All covariates from Table 3 are part of the empiri-

cal model. Robust standard errors are clustered at

the individual level in parentheses; + p < 0.10, ∗

p < 0.05, ∗∗ p < 0.01.

Table 4: Time Preferences and GDR: No Signs of Convergence

One might wonder whether unobservables that affect the distribution of time preferences

(such as East-West migration before 1961) lead to omitted variable bias. To address this

concern, we estimate bounds for our GDR treatment effect from Table 3 based on the approach

suggested by Oster (forthcoming). The crucial assumption of Oster’s approach is that selection

on observables is informative about selection on unobservables and that we can learn something

from the change of coefficients when we add additional observable variables to a regression of

the outcome on the treatment indicator. If we are furthermore willing to assume (1) that

observable and unobservable covariates are equally correlated with the GDR treatment and (2)

that the R2 of the hypothetical regression of our outcomes on GDR treatment as well as both

observed and unobserved controls is 1.3 times the R2 of our preferred specification, we can

calculate bounds for the treatment effect. Applying Oster’s formulas to the problem at hand,

we obtain the following bounding set for the treatment effect: [−0.203,−0.197]. Since the
estimated bounds do not include zero, the result suggests robustness against omitted variable

bias.

18



4.2 Results from Geographic Regression Discontinuity Design

To best communicate our Geographic Regression Discontinuity Design results, we include graph-

ical illustrations of the potential GDR treatment effects. Figures 2-3 show sample averages of

the outcome variables within bins, fourth-order polynomials fitted separately on both sides of

the FRG/GDR border as well as the optimal bandwidths according to Imbens and Kalyanara-

man (2012), which are also used in our GRDD estimations (dashed vertical lines). The plots

reveal a clear discontinuity of present bias at the erstwhile border. In contrast, the plot for

patience does not suggest a systematic change at the border. This is in accordance with our

previous results.

4
4

.5
5

5
.5

-200 -100 0 100 200
distance to border (in km)

Sample average within bin Polynomial fit of order 4

GRDD plot: Survey measure present bias

Figure 2: GRDD Plot for Present Bias (where negative numbers refer to distance of FRG

counties to the border and dashed vertical lines represent optimal bandwidth according to

Imbens and Kalyanaraman (2012))

19



5
.5

6
6

.5
7

7
.5

-200 -100 0 100 200
distance to border (in km)

Sample average within bin Polynomial fit of order 4

GRDD plot: Survey measure patience

Figure 3: GRDD Plot for Patience (where negative numbers refer to distance of FRG counties

to the border and dashed vertical lines represent optimal bandwidth according to Imbens and

Kalyanaraman (2012))

The GRDD plots provide a clear discontinuity only for present bias. Hence, in Table 5,

we present the results of our various GRDD specifications only for present bias. Overall, we

find a GDR treatment effect relatively similar to the one documented in Section 4.1 across all

specifications. However, the magnitude of the estimated GDR treatment effect is somewhat

larger in comparison to the ordinary least squares estimates. Possibly, living closer to the

FRG/GDR border strengthens the GDR treatment effect. The donut specifications are meant

to address issues of non-random selection along the border, and their results are clearly in line

with our main result.

20



(1) (2) (3)

Bandwidth h∗ 1.25h∗ 0.75h∗

Main Results

GDR -0.461 -0.401 -0.550

(0.091)∗∗ (0.081)∗∗ (0.107)∗∗

[0.106]∗∗ [0.097]∗∗ [0.124]∗∗

N 14,888 19,367 11,265

R2 0.006 0.005 0.008

Donut Results

GDR -0.379 -0.298

(0.130)∗∗ (0.105)∗∗

[0.137]∗∗ [0.114]∗∗

N 12,963 17,442

R2 0.003 0.003

Notes: We use SOEP data from 2008 and 2013. Param-

eter estimates are based on Specification 1 detailed in

Section 3.2.2. The dependent variable is the present bias

measure. Robust standard errors clustered at the indi-

vidual (county) level are reported in parentheses (square

brackets); + p < 0.10, ∗ p < 0.05, ∗∗ p < 0.01. The donut

specification relies on observations within a bandwidth

of h∗ or 1.25h∗ around the border excluding observations

closer than 40 km (10 km) on the Western (Eastern) side

of the border.

Table 5: Present Bias and GDR: Geographic Regression Discontinuity Design

4.3 Exploring Variation in Treatment Intensity: Duration of Regime

Exposure

We consider the effect of the number of years spent under socialism on present bias by defining

four groups according to respondent year of birth and running regressions for each cohort

(Columns (1)-(4) in Table 6). The results from our ordinary least squares regressions in the

upper half of the table support the simple intuition that both having spent a long period of time

under the GDR regime and having received indoctrination during daycare, school, and youth

organization attendance entailed important effects on the present bias of a regime’s citizens. In

light of our discussion of the malleability of preferences in Section 3.1, it is important to note

that we can be sure that the relevant, formative period of life was experienced subject to the

21



GDR regime only with respect to the second cohort, but also – with due qualifications – with

respect to the first and third cohort. With respect to the youngest cohort in Column (4), our

result from the ordinary least squares regression is positive, small, and insignificant. In our

GRDD results, we find a hump-shaped pattern when looking at coefficient magnitudes, implying

a greater present bias regime effect on individuals with more time under regime policies and

those with an education that was begun and completed under the regime. This is consistent

with the argumentation concerning the formation of time preferences presented in Section 3.1.

(1) (2) (3) (4)

Birth year <46 Birth year 46-60 Birth year 61-75 Birth year 76-89

Ordinary Least Squares Regressions

GDR -0.209∗∗ -0.236∗∗ -0.185∗∗ 0.014

(0.058) (0.057) (0.060) (0.068)

N 8,815 9,270 9,085 5,939

R2 0.025 0.011 0.017 0.017

Geographic Regression Discontinuity Design

GDR -0.454 -0.458 -0.661 -0.522

(0.184)∗ (0.163)∗∗ (0.173)∗∗ (0.236)∗

[0.205]∗ [0.156]∗∗ [0.189]∗∗ [0.225]∗

N 3,950 4,253 4,423 1,863

R2 0.011 0.008 0.014 0.017

Notes: We use SOEP data from 2008 and 2013. The dependent variable is the SOEP measure for

present bias. OLS: All covariates from Table 3 are part of the empirical model. Robust standard

errors are clustered at the individual level. GRDD: Robust standard errors are clustered at the

individual (county) level in parentheses (square brackets); + p < 0.10, ∗ p < 0.05, ∗∗ p < 0.01.

Table 6: Present Bias and GDR Treatment: Cohorts Results

The prior literature has reported similar findings for other economic preferences. For exam-

ple, Alesina and Fuchs-Schündeln (2007) report that greater support for a government role is

more pronounced the longer the subject has lived under the socialist regime. Similarly, Heineck

and Süssmuth (2013) find that any divergence between former GDR and FRG residents with

respect to risk preference is more pronounced for older individuals.

4.4 Robustness Checks

In this section, we provide a series of robustness checks to demonstrate that our main results

are not affected by further observed heterogeneity, measurement issues, or modeling choices.

Differences During the Treatment Period We suggest that the political regime differ-

ences contributed to the observed divergence in present bias. These regime differences had

22



substantial consequences for the economic development in both countries during the treatment

period. In an attempt to take these economic differences into account, we conduct two analyses.

First, we include the average GDP growth rate during the impressionable years of members of

both treatment and control groups in our ordinary least squares regressions.22 The basic idea is

that time preferences are formed by average GDP growth experience in early adulthood, since

a more or less prosperous economic environment might impact on the extent of the individual

bias for the present.23 Following the literature, we define these distinctive years as the period

between 18 and 25 years of age and restrict our working sample to respondents who start with

their impressionable years during the treatment period. Note that the applied age range is

consistent with the period in which time preferences are malleable according to our working

sample (Section 3.1). The ordinary least squares results in the upper part of Table 7 show that

the coefficient of the GDR treatment dummy is significant and negative, and that the extent

of present bias is smaller for former GDR citizens when higher average GDP growth rates were

experienced during their impressionable years. Second, we consider a proxy for heterogeneous

wealth during the treatment period. In particular, we include a dummy variable of house own-

ership status in 1990 before reunification in our regression exercises.24 The results in the lower

part of Table 7 reveal that we find a significant negative treatment effect again and that house

ownership has differential effects on present bias.

22The impressionable years hypothesis proposes that individuals are susceptible to attitude change during
late adolescence and early adulthood and that susceptibility drops radically thereafter. Evidence confirming the
hypothesis is presented, for example, in Krosnick and Alwin (1989). Giuliano and Spilimbergo (2014) provide
a recent application in economics. The GDP data for the years 1947 to 1989 for East and West Germany and
from 1990 to 1991 for West Germany are from Ritschl and Spoerer (1997). The GDP data from 1990 to 1991 for
East Germany are from Sleifer (2006). These data are available at: histat.gesis.org/histat/. The data for
the period 1992 to 2017 stem from the German Statistical Offices (see statistik-bw.de/VGRdL/tbls/?lang=
en-GB.

23Note that we cannot use unemployment as an economic indicator, since unemployment was officially equal
to zero in the former GDR.

24As a qualifier, we attest that home ownership may be a bad control variable. At least in the GDR, home
ownership was related to the individual willingness to achieve this end, which might have been influenced by
the level of self-control.

23

histat.gesis.org/histat/
statistik-bw.de/VGRdL/tbls/?lang=en-GB
statistik-bw.de/VGRdL/tbls/?lang=en-GB


Average GDP Growth Rate

GDR -0.1341∗

(0.0553)

Average GDP Growth Rate -0.0160

(0.0104)

Average GDP Growth Rate × GDR -0.0446∗

(0.0196)

N 31,568

R2 0.0124

Home Ownership

GDR -0.3315∗∗

(0.0798)

Home Ownership -0.1462+

(0.0765)

Home Ownership × GDR 0.2089+

(0.1231)

N 7,407

R2 0.0148

Notes: We use SOEP data from 2008 and 2013. Pa-

rameter estimates from ordinary least squares regres-

sions. The dependent variable is the SOEP measure

for present bias. All covariates from Table 3 are part

of the empirical model. Robust standard errors are

clustered at the individual level in parentheses; +

p < 0.10, ∗ p < 0.05, ∗∗ p < 0.01.

Table 7: Present Bias and GDR Treatment: Taking Pre-Reunification Differences Into Account

Differences After the Treatment Period We rely on measurements of time preferences

from 2008 and 2013. The two parts of Germany were reunified in 1990. Attributing any

difference in time preferences to the GDR’s socialist regime presupposes that the life experiences

of East and West Germans after 1990 were either comparable, observably and measurably

different, or had insignificant effect on time preferences. One potentially important reason why

former GDR and FRG residents had different post-reunification experiences is that the post-

reunification economic performance of East Germany dramatically lagged West Germany’s. In

particular, this becomes manifest in the dramatic increase in the official unemployment rate in

East Germany up to roughly 20% after reunification. To test whether the economic turmoil

after reunification had an impact on time preferences measured in 2008 and 2013, we use the

salient unemployment experience of former GDR citizens compared to FRG citizens as an

indicator. We conduct a linear mediation analysis with the outcome present bias, in which we

attempt to disentangle direct regime effects on present bias and “indirect” effects via different

24



unemployment trajectories in West and East Germany, the latter at least partly triggered by

the GDR treatment. In particular, we use the following system of linear equations:

unempexpit =α1 + β1GDRi + ω1xit + �it,1, (2)

present biasit =α2 + β2GDRi + γunempexpit + νunempexpit ×GDRi + ω2xit + �it,2, (3)

where unempexpit stands for years of unemployment experienced by individual i (resulting

from an involuntary job loss) at time t and xit stands for the vector of covariates also used

in the ordinary least squares regressions above. The average mediation effect for former GDR

citizens is given by δ = β1(γ + ν), indicating the indirect regime effect on the individual extent

of present bias of former GDR citizens transmitted via the own unemployment experience after

reunification. The average direct effect is given by ζ = β2 + ν[α1 + β1 + ω1E(xi)], indicating the

direct effect of the GDR’s political regime on present bias given the years of unemployment.

To disentangle the total causal effect into the two causal channels defined above, two crucial

assumptions must hold (e.g., Huber 2014, Imai et al. 2010 a,b): (1) conditional independence

of the GDR treatment, that is, there are no unobserved confounders jointly affecting both the

GDR treatment and the years of unemployment experience and/or time preferences conditional

on the vector of covariates. (2) conditional independence of the mediator, that is, conditional on

GDR treatment and the vector of covariates, unemployment experience is effectively randomly

assigned. With respect to assumption (1), we acknowledge again that the German history of

separation and reunification represents a natural experiment. To ensure assumption (2) in our

empirical analysis below, we compare only those members of the control and the treatment

group who have either experienced no unemployment spell at all or whose first job loss after

the German reunification was due to a plant closure. The main argument is that plant closures

are an arguably exogenous event from the perspective of the individual worker (Anger et al.

2017, Marcus 2013), in particular for former GDR citizens who worked for former state-owned

companies.

Table 8 shows the estimated average direct and indirect effects of the GDR treatment on

present bias for the treatment group of former GDR citizens. Column (1) presents the results

for the working sample described above, Column (2) documents the results for a subgroup

of the working sample who spent all of their impressionable years either in the GDR or the

FRG. The direct effect of the GDR treatment on present bias is significantly negative in both

specifications, supporting our main results. The size of the estimated indirect effect is small

and significant in Column (2) (but only at the ten percent level). In this case, the ratio of the

indirect and the direct effects amounts to roughly ten percent, indicating that the direct regime

effect remains important when we try to take indirect, yet salient post-reunification experiences

25



into account.

(1) (2)

Full Sample Impressionable years sample

Direct effect -0.1923∗∗ -0.2405∗∗

(0.0423) (0.0545)

Indirect effect 0.0119 0.0204+

(0.0079) (0.0123)

Ratio indirect/direct -0.062 -0.085+

(0.0405) (0.0502)

N 18,654 11,436

Notes: We use SOEP data from 2008 and 2013. In Column (2), we focus

on subjects who spent all of their impressionable years either in the GDR or

the FRG. Estimated direct and indirect effects are calculated from parameter

estimates of equation system (2) and (3) based on linear structural equation

modeling. Robust standard errors are clustered at the individual level. z

statistics in parentheses; + p < 0.10, ∗ p < 0.05, ∗∗ p < 0.01.

Table 8: Present Bias, Unemployment, and GDR Treatment: Results from a Mediation Analysis

In addition to the mediation analysis using unemployment, we consider an identifiable sub-

group who had a relatively “smooth transition” in economic terms, that is, individuals who –

during GDR times – had a social network with links into the FRG.25 This subgroup is likely to

have had relatively better post-reunification experiences based on the argument that Burchardi

and Hassan (2013) establish for West Germans with social ties in East Germany. The positive

income impact of social ties that Burchardi and Hassan report can also be found in our data

for East Germans with links in the FRG. Focusing on this subpopulation of East Germans, we

find that they also display a lower present bias. The coefficient using ordinary least squares is

−0.26∗∗, while it is about −0.33∗∗ when using the GRDD approach.26

Migration During the Treatment Period There was significant migration from East to

West Germany after the treatment imposition and before the erection of the Berlin Wall, which

is potentially problematic for our identification. For subjects in our sample who participated

in at least in one of the survey years 1990 to 1993, we can reconstruct whether or not they

migrated from East to West Germany during the separation period. Applying our identification

strategies to the resulting subsample, we obtain results that are consistent with our main results

(see Tables 12-13 in Appendix B).

25The information about social ties stems from a question addressed at participants of the new East German
SOEP sample in spring 1990. At the household level, participants were asked whether they have had relatives
or friends in the former FRG.

26Regression results are available upon request.

26



Migration After the Treatment Period After 1989, notable migration from East to West

Germany, and some movement from West to East Germany took place (e.g., Hunt 2006, Fuchs-

Schündeln and Schündeln 2009). The SOEP provides us with information on both directions of

migration between West and East Germany after reunification. When we add dummy variables

indicating whether someone has moved from East Germany to West Germany or vice versa in

our ordinary least squares approach, the estimated treatment effect is −0.21∗∗ [s.e. : 0.03] and
the two ordinary least squares coefficients of the migration dummy variables are not significantly

different from zero at α = 0.05.27

Measurement Of Time Preferences In Section 3.1, we documented that present bias

in particular is remarkably stable in middle age, while we found notable variation in early

adulthood and old age. This explains why we use all available time preferences information in

our regression exercises. However, it is interesting to check whether our results are robust to

considering only one observation per person. To this end, we consider (1) the average value

of measurements of present bias and (2) only the latest available information on present bias.

With respect to the average value, we obtain estimated treatment effects of −0.22∗∗ [s.e. : 0.03]
using ordinary least squares and significantly negative GDR treatment effects in the range

[−0.70∗∗,−0.48∗∗] in our GRDD approach depending on the chosen bandwidth. Considering the
latest available information, we find a significantly negative GDR treatment effect on present

bias of −0.20∗∗ [s.e. : 0.03] applying ordinary least squares and significantly negative GDR
treatment effects in the range of [−0.72∗∗,−0.48∗∗] using our GRDD specification.

Regional Effects We argue that we are capturing the influence of GDR’s socialist regime.

It may be conjectured that the estimated average treatment effects mask heterogeneity across

GDR regions. To explore this aspect, we consider specifications that comprise state-specific

effects for each state from East Germany (applying the post-reunification demarcation of states

that was almost identical to that between 1945 to 1952) and a dummy variable indicating 1

for survey respondents without state information. Our results indicate some (albeit no con-

siderable) heterogeneity across states (see Table 14 in Appendix B). One interpretation of the

limited heterogeneity across states is that it is due to limited regional variation in treatment

intensity. Many, if not most critical characteristics of the GDR regime (e.g., the indoctrination

in educational facilities) applied rather uniformly to all regions. In terms of regional effects,

we may also take into account that East and West Germany were and continue to be very

27Remember that, in our GRDD estimations, we exclude survey respondents who moved from West to East
or from East to West Germany within the relevant post-reunification period.

27



different in terms of religion.28 This itself may be an outcome of the political regime differences

in East and West Germany (since religiosity was suppressed in the GDR). Before treatment

imposition, residents who lived in the area of what became the GDR were mostly Protestant,

whereas residents of what became the FRG were either Protestant or Catholic, as explained by

Bauernschuster and Rainer (2012). We restrict our sample to regions that were unambiguously

Protestant – that is, we omit data from Bavaria, Baden-Wuerttemberg, Northrhine-Westfalia,

Rhineland-Palatinate, and Saarland – and find that our results from Section 4.1 are robust to

this consideration.29

Influence on Related Personality Traits Economic preferences and personality traits are

considered to be complements when explaining important life outcomes and behavior (e.g.,

Humphries and Kosse 2017). Nevertheless, there are some conceptual relationships. For ex-

ample, Almlund et al. (2011: p. 70) hypothesize that time preferences and conscientiousness

are most likely related. Becker et al. (2012) find a small correlation in a large SOEP data set

between time preferences and conscientiousness and a medium-sized one between time prefer-

ences and agreeableness.30 Based on this argumentation and results, it is interesting to explore

whether conscientiousness and agreeableness are also influenced by the GDR’s regime. We

build on and extend Friehe et al. (2015) for this purpose.31 Table 9 presents the results using

our first identification strategy.

28Religion may be relevant to time preferences. For instance, Doepke and Zilibotti (2014) argue that religious
beliefs such as Protestantism can be seen as complementary drivers of patience and work ethic.

29Regression results are available upon request.
30The psychological literature defines conscientiousness as the tendency to be organized, responsible, and hard

working and agreeableness is defined as the tendency to act in a cooperative, unselfish manner (e.g., Almlund
et al. 2011: p. 11).

31The SOEP comprises information about conscientiousness and agreeableness in 2005 and 2009 in the form of
respondents’ self-positioning on scales ranging from 1 to 7 as an answer to how well specific statements describe
their personality. We generate measures of agreeableness and conscientiousness by standardizing the sum of the
scores of the dimension-specific questions whereby higher values indicate a stronger intensity of the particular
trait (e.g. being more agreeable).

28



(1) (2)

Conscient- Agree-

iousness ableness

GDR 0.0984∗∗ 0.0285∗

(7.29) (2.09)

N 35,417 35,536

R2 0.048 0.039

Notes: We use SOEP data from 2005 and 2009. Pa-

rameter estimates are from ordinary least squares re-

gressions. The dependent variable is the respective

standardized personality measure. The specification

includes all covariates from Table 3. Robust stan-

dard errors are clustered at the individual level. t

statistics in parentheses; + p < 0.10, ∗ p < 0.05, ∗∗

p < 0.01.

Table 9: GDR Treatment Effects on Conscientiousness and Agreeableness: Results Using Or-

dinary Least Squares Regressions

The estimated GDR treatment effect indicates that former GDR citizens have a higher

conscientiousness score than former FRG citizens. Since a higher degree of conscientiousness

is usually associated with higher self-control, the result is consistent with the observed lower

present bias of former GDR residents (i.e., our main result). The GDR treatment effect is also

significant and positive for the trait agreeableness. Gill and Prowse (2016), for example, provide

evidence that more agreeable players learn faster and perform better in strategic interactions.

Most likely, better performance in strategic interactions is positively linked to better self-control

and therefore to less present bias.

When we turn to the second identification strategy, we find that the GRDD plots do not

reveal a clear discontinuity at the former border between the FRG and the GDR with respect

to both personality measures. In view of the small correlations between time preferences and

these two personality traits reported in Becker et al. (2012), this outcome does not surprise.

29



-.
2

0
.2

.4
.6

-200 -100 0 100 200
distance to border (in km)

Sample average within bin Polynomial fit of order 3

GRDD plot: Conscientiousness

-.
3

-.
2

-.
1

0
.1

.2

-200 -100 0 100 200
distance to border (in km)

Sample average within bin Polynomial fit of order 3

GRDD plot: Agreeableness

Figure 4: GRDD Plots for Conscientiousness and Agreeableness (where negative numbers refer

to distance of FRG counties to the border and dashed vertical lines represent optimal bandwidth

according to Imbens and Kalyanaraman (2012))

Probit Regression In Table 2, we presented descriptive evidence using a dummy variable

that is equal to one when the answer to the question referring to present bias was higher than

six. Using this information as the dependent variable in a probit regression yields an average

partial effect for the GDR regime amounting to −0.048∗∗. To offer an interpretation of our
result, the probability that an individual shows present bias is reduced by 17.5% when the

individual has experienced the GDR regime instead of the FRG.

5 Exploring Implications of GDR Experience

Both of our identification strategies have shown that native Germans who lived in the GDR

in 1989 (still) differ from those who lived in the FRG in terms of present bias. In this section,

we explore some potential economic consequences of this long-lasting impact of the GDR’s

political regime, using outcome measures from the SOEP that are commonly understood to

be determined by present bias (e.g., Meier and Sprenger 2010, O’Donoghue and Rabin 2006).

Specifically, we will test whether our survey measures of present bias predict health related

outcomes (body mass index, obesity, and smoking); savings and credit uptake behavior; and

education outcomes (specifically, whether or not the individual obtained a college or a university

degree) while controlling for GDR treatment status.

We report two sets of results in Table 10. First, we use two measures of present bias in pooled

ordinary least squares and pooled Probit regression exercises to analyze the association between

present bias and selected economic outcomes. Second, we use the lagged value of present bias

reported in 2008 as an instrument (as exclusion restriction) in various econometric specifications

in an attempt to address the possibility of unobserved confounding between present bias and our

selected economic outcomes. In particular, we use a two-stage-least-square approach, a control

30



function approach, and a recursive two-equations system depending on the nature of both the

outcome and the present bias variable. In all econometric specifications, we are interested in

the predictive power of present bias on the economic outcome of interest. Information on the

outcome measures in the upper part of Table 10 stem from 2008 and 2013 for smoking, saving,

credit, and education, and from 2008 and 2014 for BMI and obesity. In the lower part of Table

9, information on the economic outcomes stem from 2013 and 2014.

In line with the literature, we find that our survey measures of present bias have predictive

power for all outcomes considered in almost all cases. A greater present bias is associated with

a larger BMI, increased likelihood of being obese, greater propensity to be a smoker, decreased

probability of saving regularly (only in the pooled specifications), raised probability of having to

pay back a consumer credit, and a lowered chance of obtaining a college or university degree. For

example, being present biased (survey response > 6) correlates to an increase in the probability

to smoke by about five to fifteen percentage points, depending on the econometric specification.

Compared to the raw mean of smoking status in our working sample, this implies a notable

change of 20% to 58%. Furthermore, we observe significant differences between the control and

the treatment group for all outcomes. To illustrate the importance of the GDR treatment effect

on the outcomes of interest relative to the effect of present bias, let us consider the probability

of having to pay back a credit: Being a former citizen of the GDR implies an increase in the

probability of an outstanding credit obligation by roughly 7 percentage points, while present

bias alone correlates with an increase in the same credit obligation probability by about 2 to 8

percentage points depending on the specification. Hence, when taken at face value, the effect

of having lived in the GDR is quantitatively at least as important for the prediction of credit

behavior as present bias, allowing for the fact that, on average, former GDR citizens are less

present biased than former FRG citizens.

6 Conclusion

Time preferences are integral to modern economics. They are critical to a host of decisions,

many of which have key impacts on important life outcomes (Golsteyn et al. 2014). There

is considerable heterogeneity regarding time preferences and relatively little knowledge about

the sources of this variety (Falk et al. forthcoming, Wang et al. 2016). Intergenerational

transmission and the institutional environment likely play important roles. In this line of

inquiry, the present study contributes by exploring whether the GDR’s socialist institutional

environment has left a notable and lasting footprint on its former citizens’ time preferences.

31



B
o
d

y
M

a
ss

In
d

e
x

O
b

e
si

ty
S

m
o
k
in

g
S

a
v
in

g
C

re
d

it
E

d
u

c
a
ti

o
n

O
L

S
P

ro
b

it
P

ro
b

it
P

ro
b

it
P

ro
b

it
P

ro
b

it
G

D
R

0
.4

8
9
*
*

0
.4

9
0
*
*

0
.0

2
8
*
*

0
.0

2
8
*
*

0
.0

3
0
*
*

0
.0

3
0
*
*

-0
.0

5
3
*
*

-0
.0

5
3
*
*

0
.0

7
2
*
*

0
.0

7
2
*
*

-0
.1

2
0
*
*

-0
.1

2
1
*
*

(0
.0

7
3
)

(0
.0

7
3
)

(0
.0

0
6
)

(0
.0

0
6
)

(0
.0

0
6
)

(0
.0

0
6
)

(0
.0

0
7
)

(0
.0

0
7
)

(0
.0

0
5
)

(0
.0

0
5
)

(0
.0

1
1
)

(0
.0

1
1
)

P
re

se
n
t

b
ia

s
0
.0

6
9
*
*

0
.0

0
4
*
*

0
.0

1
2
*
*

-0
.0

0
6
*
*

0
.0

0
7
*
*

-0
.0

1
2
*
*

(0
.0

1
4
)

(.
0
0
1
)

(.
0
0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
2
)

D
u

m
m

y
p

re
se

n
t

b
ia

s
0
.3

3
1
*
*

0
.0

2
0
*
*

0
.0

4
3
*
*

-0
.0

2
4
*
*

0
.0

2
1
*
*

-0
.0

5
0
*
*

(0
.0

6
4
)

(0
.0

0
5
)

(0
.0

0
5
)

(0
.0

0
6
)

(0
.0

0
5
)

(0
.0

1
0
)

F
-T

e
st

/
L

R
-T

e
st

(d
f=

2
2
)

1
0
7
.1

0
*
*

1
0
7
.1

1
*
*

5
8
6
.2

9
*
*

5
8
5
.6

2
*
*

1
4
7
0
.9

1
*
*

1
4
5
9
.1

6
*
*

5
3
3
.7

8
*
*

5
2
8
.9

0
*
*

1
4
9
6
.1

8
*
*

1
4
9
2
.0

6
*
*

9
4
6
.4

9
*
*

9
4
1
.7

0
*
*

N
3
1
,2

1
0

3
1
,2

1
0

3
1
,5

1
6

3
2
,7

0
5

3
2
,6

3
5

7
,6

4
6

A
ll
o
w

in
g

fo
r

u
n

o
b

se
re

v
d

c
o
n

fo
u

n
d

e
rs

:
In

st
ru

m
e
n
t/

e
x
c
lu

si
o
n

re
st

ri
c
ti

o
n

:
la

g
g
e
d

p
re

se
n
t

b
ia

s
(t

-5
)

2
S

L
S

2
S

L
S

P
ro

b
it

-C
F

P
ro

b
it

-R
S

P
ro

b
it

-C
F

P
ro

b
it

-R
S

P
ro

b
it

-C
F

P
ro

b
it

-R
S

P
ro

b
it

-C
F

P
ro

b
it

-R
S

P
ro

b
it

-C
F

P
ro

b
it

-R
S

G
D

R
0
.5

0
1
*
*

0
.5

3
6
*
*

0
.0

2
9
*
*

0
.0

3
1
*
*

0
.0

2
7
*
*

0
.0

3
0
*
*

-0
.0

3
3
*
*

-0
.0

3
4
*
*

0
.0

7
1
*
*

0
.0

7
5
*
*

-0
.1

5
7
*
*

-0
.1

5
3
*
*

(0
.1

0
5
)

(0
.1

0
7
)

(0
.0

0
9
)

(0
.0

1
0
)

(0
.0

0
9
)

(0
.0

1
0
)

(0
.0

1
0
)

(0
.0

1
0
)

(0
.0

0
8
)

(0
.0

0
9
)

(0
.0

1
9
)

(0
.0

1
8
)

P
re

se
n
t

b
ia

s
0
.2

1
3
*
*

0
.0

0
9
*
*

0
.0

3
0
*
*

-0
.0

0
6

0
.0

1
3
*
*

-0
.0

3
9
*
*

(0
.0

5
7
)

(0
.0

0
5
)

(0
.0

0
4
)

(0
.0

0
5
)

(0
.0

0
4
)

(0
.0

1
0
)

D
u

m
m

y
p

re
se

n
t

b
ia

s
1
.4

8
5
*
*

0
.0

7
9
*

0
.1

4
7
*
*

-0
.0

2
8

0
.0

7
7
*

-0
.2

2
2
*
*

(0
.3

6
4
)

(0
.0

3
3
)

(0
.0

2
9
)

(0
.0

3
2
)

(0
.0

3
0
)

(0
.0

5
7
)

F
-T

e
st

/
L

R
-T

e
st

(d
f)

3
4
.4

3
*
*

3
3
.8

1
*
*

2
1
3
.2

8
*
*

1
1
6
8
.0

4
*
*

5
8
4
.0

4
*
*

1
5
8
3
.1

6
*
*

1
6
5
.3

2
*
*

1
1
4
2
.8

8
*
*

6
5
7
.3

1
*
*

1
6
5
6
.7

4
*
*

3
6
9
.5

2
*
*

6
3
4
.1

0
*
*

(2
2
)

(2
2
)

(2
2
)

(4
4
)

(2
2
)

(4
4
)

(2
2
)

(4
4
)

(2
2
)

(4
4
)

(2
2
)

(4
4
)

N
9
,6

3
9

9
,6

3
9

9
,7

6
4

1
0
,4

6
8

1
0
,4

4
0

2
,2

6
0

N
o
te
s
:

W
e

u
se

S
O

E
P

d
a
ta

fr
o
m

2
0
0
8
,

2
0
1
3
,

a
n

d
2
0
1
4
.

O
u

tc
o
m

e
m

ea
su

re
s

in
th

e
u

p
p

er
p

a
rt

o
f

th
e

ta
b

le
st

em
fr

o
m

ei
th

er
2
0
0
8

a
n

d
2
0
1
4

(B
M

I,
O

b
es

it
y
)

o
r

2
0
0
8

a
n

d
2
0
1
3

(S
m

o
k
in

g
,

S
a
v
in

g
,

C
re

d
it

,
E

d
u

ca
ti

o
n

)
a
n

d
in

th
e

lo
w

er
p

a
rt

o
f

th
e

ta
b

le
fr

o
m

ei
th

er
2
0
1
3

o
r

2
0
1
4
.

B
o
d

y
M

a
ss

In
d

ex
:

B
M

I-
fo

rm
u

la
;

O
b

es
it

y
:

D
u

m
m

y
v
a
ri

a
b

le
is

eq
u

a
l

to
1

if
B

M
I
≥

3
0
;

S
m

o
k
in

g
:

D
u

m
m

y
v
a
ri

a
b

le
is

eq
u

a
l

to
1

if
“
y
es

”
.

S
a
v
in

g
:

D
u

m
m

y
v
a
ri

a
b

le
is

eq
u

a
l

to
1

if
re

sp
o
n

d
en

ts
sa

v
e

so
m

e
m

o
n

ey
ev

er
y

m
o
n
th

.
C

re
d

it
:

D
u
m

m
y

v
a
ri

a
b

le
is

eq
u

a
l

to
1

if
re

sp
o
n

d
en

ts
p

a
y

b
a
ck

a
co

n
su

m
er

cr
ed

it
.

E
d

u
ca

ti
o
n

:
D

u
m

m
y

v
a
ri

a
b

le
is

eq
u

a
l

to
1

if
re

sp
o
n

d
en

ts
h

a
v
e

a
co

ll
eg

e
o
r

u
n

iv
er

si
ty

d
eg

re
e.

T
h

e
d

u
m

m
y

v
a
ri

a
b

le
fo

r
p

re
se

n
t

b
ia

s
is

eq
u

a
l

to
o
n

e
w

h
en

th
e

re
sp

o
n

se
w

a
s

h
ig

h
er

th
a
n

si
x

o
n

a
te

n
-i

te
m

sc
a
le

.
T

h
e

sp
ec

ifi
ca

ti
o
n

s
in

cl
u

d
e

a
ll

co
v
a
ri

a
te

s
fr

o
m

T
a
b

le
3
.

A
v
er

a
g
e

p
a
rt

ia
l

eff
ec

ts
(A

P
E

)
a
re

p
re

se
n
te

d
.

P
ro

b
it

-C
F

:
C

o
n
tr

o
l-

fu
n

ct
io

n
a
p

p
ro

a
ch

w
it

h
la

g
g
ed

p
re

se
n
t

b
ia

s
(t

-5
)

a
s

ex
cl

u
si

o
n

re
st

ri
ct

io
n

.
P

ro
b

it
-R

S
:

R
ec

u
rs

iv
e

p
ro

b
it

tw
o
-e

q
u

a
ti

o
n

sy
st

em
w

it
h

la
g
g
ed

p
re

se
n
t

b
ia

s
(t

-5
)

a
s

ex
cl

u
si

o
n

re
st

ri
ct

io
n

.
B

o
th

p
a
rt

s
o
f

th
e

ta
b

le
:

C
o
lu

m
n

s
(1

)-
(2

):
F

-T
es

t
o
v
er

a
ll

si
g
n

ifi
ca

n
ce

;
C

o
lu

m
n

s
(3

)-
(1

2
):

L
R

-T
es

t
o
v
er

a
ll

si
g
n

ifi
ca

n
ce

.
A

d
d

it
io

n
a
l

in
fo

rm
a
ti

o
n

o
n

th
e

lo
w

er
p

a
rt

o
f

th
e

ta
b

le
:

C
o
lu

m
n

s
(1

)-
(2

):
F

-T
es

t
ex

cl
u

d
ed

in
st

ru
m

en
t

fi
rs

t
st

a
g
e:

1
6
0
6
.2

6
∗∗

;
8
1
4
.6
∗∗

.
C

o
lu

m
n

s
(3

),
(5

),
(7

),
(9

),
(1

1
):

W
a
ld

te
st

o
f

ex
o
g
en

ei
ty

(d
f=

1
):

0
.3

0
,

2
4
.5

2
∗∗

,
0
.3

7
,

1
.1

9
,

4
.7

5
∗
.

C
o
lu

m
n

s
(4

),
(6

),
(8

),
(1

0
),

(1
2
):

L
R

-t
es

t
o
f

rh
o
=

0
(d

f=
1
):

3
.5

1
+

,
1
3
.6

6
∗∗

,
0
.0

2
,

3
.7

5
+

,
6
.2

7
∗
.

S
ta

n
d

a
rd

er
ro

rs
a
re

cl
u

st
er

ed
a
t

th
e

in
d

iv
id

u
a
l

le
v
el

if
p

o
o
le

d
d

a
ta

a
re

em
p

lo
y
ed

a
n

d
p
re

se
n
te

d
in

p
a
re

n
th

es
es

.
P

ro
b

it
-R

S
:

B
o
o
ts

tr
a
p

p
ed

st
a
n

d
a
rd

er
ro

rs
w

it
h

5
0
0

re
p

li
ca

ti
o
n

s.
+
p
<

0
.1

0
,
∗
p
<

0
.0

5
,
∗∗

p
<

0
.0

1
.

T
a
b
le

1
0
:

P
re

se
n
t

B
ia

s
a
n
d

D
ir

ec
t

a
n
d

In
d
ir

ec
t

T
re

a
tm

en
t

E
ff

ec
ts

32



Our results suggest that – indeed – the GDR’s socialist regime significantly and enduringly

diminished the present bias of its former citizens. We find survey-based evidence indicating

influence of political institutions on time preferences. Significant differences in East and West

Germans’ levels of present bias remain two decades after reunification and the data does not

suggest convergence.

The reported level of patience of former GDR citizens is comparable to that of former FRG

citizens. The relatively small variation is directionally aligned with our hypothesis about higher

patience among former GRD residents (but insufficient to generate significant differences) and

consistent with small variation in patience levels across much more diverse countries (Wang et

al. 2016).

Acknowledgments

We gratefully acknowledge the very helpful comments made by Daniel Harenberg, Mario Mech-

tel, Hannah Schildberg-Hörisch, Georg Weizsäcker, and participants of the 2017 German Eco-

nomic Association Meeting in Vienna on earlier versions of the paper. We are also very thankful

for the insightful suggestions received from three anonymous reviewers.

References

Acemoglu, D., Johnson, S., and J.A. Robinson, 2001. The colonial origins of comparative de-

velopment: An empirical investigation. American Economic Review 91, 1369-1401.

Alesina, A., and N. Fuchs-Schündeln, 2007. Good-bye Lenin (or not?): The effect of Commu-

nism on People’s Preferences. American Economic Review 97, 1507-1528.

Almlund, M., Duckworth, A., Heckman, J., and T. Kautz, 2011. Personality psychology and

economics. In: Hanushek, E.A., Machin, S., Woessmann, L. (Eds.), Handbook of the economics

of education. Amsterdam: North-Holland.

Ameriks, J., Caplin, A., Leahy, J., and T. Tyler, 2007. Measuring self-control problems. Amer-

ican Economic Review 97, 966-972.

Anger, S., Camehl, G., and F. Peter, 2017. Involuntary job loss and changes in personality

traits. Journal of Economic Psychology 60, 71-91.

Anger, S., and G. Heineck, 2010. Do smart parents raise smart children? The intergenerational

transmission of cognitive abilities. Journal of Population Economics 23, 1105-1132.

Anger, S., and D.D. Schnitzlein, 2017. Cognitive skills, non-cognitive skills, and family back-

ground: evidence from siblin correlations. Journal of Population Economics 30, 591-620.

Bauernschuster, S., and H. Rainer, 2012. Political regimes and the family: how sex-role atti-

33



tudes continue to differ in reunified Germany. Journal of Population Economics 25, 5-27.

Baumeister, R., and J. Tierney, 2012. Willpower – Why self-control is the secret to success.

New York: Penguin Books.

Beblo, M., and L. Görges, 2018. On the nature of nurture. The malleability of gender differ-

ences in work preferences. Journal of Economic Behavior and Organization 151, 19-41.

Becker, S., Boeckh, K., Hainz, C., and L. Wöessmann, 2016. The empire is dead, long live the

empire! Long-run persistence of trust and corruption in the bureaucracy. Economic Journal

126, 40-74.

Becker, A., Deckers, T., Dohmen, T., Falk, A., and F. Kosse, 2012. The relationship between

economic preferences and psychological personality measures. Annual Review of Economics 4,

453-478.

Becker, G.S., and C.B. Mulligan, 1997. The endogenous determination of time preferences.

Quarterly Journal of Economics 112, 729-758.

Benhabib, J., and A. Bisin, 2011. Social construction of preferences. In: Benhabib, J., Bisin,

A., Jackson, M. (Eds.). Handbook of Social Economics 1A, 201-220.

Bettinger, E., and R. Slonim, 2007. Patience among children. Journal of Public Economics 91,

343-363.

Bishai, D.M., 2004. Does time preference change with age? Journal of Population Economics

17, 583-602.

Bisin, A., and T. Verdier, 2011. The economics of cultural transmission and socialization. In:

Benhabib, J., Bisin, A., and M.O. Jackson (Eds.). Handbook of Social Economics, Vol. 1A,

North-Holland.

Bowles, S., 1998. Endogenous preferences: The cultural consequences of markets and other

economic institutions. Journal of Economic Literature 36, 75-111.

Brenoe, A.A., and T. Epper, 2018. The intergenerational transmission of time preferences across

four decades. Working paper available at sites.google.com/view/aabrenoe/research (last

accessed August 31, 2018).

Brosig-Koch, J., Helbach, C., Ockenfels, A., and J. Weimann, 2011. Still different after all

these years: Solidarity behavior in East and West Germany. Journal of Public Economics 95,

1373-1376.

Brown, H., and M. van der Pol, 2015. Intergenerational transfer of time and risk preferences.

Journal of Economic Psychology 49, 187-204.

Bruce, G., 2010. The firm – The inside story of the Stasi. Oxford: Oxford University Press.

BStU, 2012. “Aktion Ungeziefer”. Accessed May 1 2017 at bstu.bund.de/DE/Presse/Themen/

Hintergrund/20120518_aktion_ungeziefer.html.

Bulte, E.H., and R.D. Horan, 2011. Intergenerational transmission of preferences. Economics

34

sites.google.com/view/aabrenoe/research
bstu.bund.de/DE/Presse/Themen/Hintergrund/20120518_aktion_ungeziefer.html
bstu.bund.de/DE/Presse/Themen/Hintergrund/20120518_aktion_ungeziefer.html


Letters 112, 85-87.

Burchardi, K.B., and T.A. Hassan, 2013. The economic impact of social ties: Evidence from

German reunification. Quarterly Journal of Economics 128, 1219-1271.

Burks, S., Carpenter, J., Goette, L., and A. Rustichini, 2012. Which measures of time prefer-

ences best predict outcomes: Evidence from a large-scale field experiment. Journal of Economic

Behavior and Organization 84, 308-320.

Campa, P., and M. Serafinelli, forthcoming. Politico-economic regimes and attitudes: Female

workers under state-socialism. Review of Economics & Statistics.

Carlsson, F., Johansson-Stenman, O., and P. Khanh Nam, 2014. Social preferences are stable

over long periods of time. Journal of Public Economics 117, 104-114.

Carvalho, L.S., Meier, S., and S.W. Wang, 2016. Poverty and economic decision-making: Evi-

dence from changes in financial resources at payday. American Economic Review 106, 260-284.

Chen, M.K., 2013. The effect of language on economic behavior: Evidence from savings rates,

health behaviors, and retirement assets. American Economic Review 103, 690-731.

Chowdhury, S., Sutter, M., and K.F. Zimmermann, 2018. Evaluating intergenerational persis-

tence of economic preferences: A large scale experiment with families in Bangladesh. CESifo

Working Papers 6914.

Cobb-Clark, D., and S. Schurer, 2013. Two economists’ musings on the stability of locus of

control. Economic Journal 123, F358-F400.

Cunha, F., and J.J. Heckman, 2007. The technology of skill formation. American Economic

Review Papers and Proceedings 97, 31-47.

Dasgupta, P., and E. Maskin, 2005. Uncertainty and hyperbolic discounting. American Eco-

nomic Review 95, 1290-1299.

Deckers, T., Falk, A., Kosse, F., Pinger, P., and H. Schildberg-Hörisch, 2017. Socio-economic

status and inequalities in children’s IQ and economic preferences. IZA Discussion Paper No.

11158.

Delaney, L., and O. Doyle, 2012. Socioeconomic differences in early childhood time preferences.

Journal of Economic Psychology 33, 237-247.

Dell, M., 2010. The persistent effect of Peru’s mining mita. Econometrica 78, 1863-1903.

DellaVigna, S., 2009. Psychology and economics: Evidence from the field. Journal of Economic

Literature 47, 315-372.

DellaVigna, S., and M.D. Paserman, 2005. Job search and impatience. Journal of Labor Eco-

nomics 23,527-588.

Doepke, M., and F. Zilibotti, 2008. Occupational choice and the spirit of capitalism. Quarterly

Journal of Economics 123, 747-793.

Doepke, M., and F. Zilibotti, 2014. Culture, entrepreneurship, and growth. Handbook of Eco-

35



nomic Growth 2, 1-48.

Dohmen, T., Falk, A. , Huffman, D., and U. Sunde, 2010. Are risk aversion and impatience

related to cognitive ability? American Economic Review 100, 1238-1260.

Dohmen, T., Falk, A. , Huffman, D., and U. Sunde, 2012. The intergenerational transmission

of risk and trust attitudes. Review of Economic Studies 79, 645-677.

Dohmen, T., Enke, B., Falk, A., Huffman, D., and U. Sunde, 2018. Patience and Comparative

Development. Mimeo.

Dragone, D., and N. Ziebarth, 2017. Economic development, novelty consumption, and body

weight: Evidence from the East German transition to capitalism. Journal of Health Economics

51, 41-65.

Falk, A., Becker, A., Dohmen, T., Enke, B., Huffman, D., and U. Sunde. forthcoming. Global

evidence on economic preferences. Quarterly Journal of Economics.

Fehr, E., Bernhard, H., and B. Rockenbach, 2008. Egalitarianism in young age. Nature 454,

1079-1083.

Fehr, E., and K. Hoff, 2011. Introduction: Tastes, castes and culture: The influence of society

on preferences. Economic Journal 121, F396-F412.

Fourage, D., Kriechel, B., and T. Dohmen, 2014. Occupational sorting of school graduates: The

role of economic preferences. Journal of Economic Behavior and Organization 106, 335-351.

Friehe, T., and M. Mechtel, 2014. Conspicuous consumption and political regimes: Evidence

from East and West Germany. European Economic Review 67, 62-81.

Friehe, T., Pannenberg, M., and M. Wedow, 2015. Let bygones be bygones? Socialist regimes

and personalities in Germany. SOEPpapers 776.

Fuchs-Schündeln, N., and T.A. Hassan, 2015. Natural experiments in macroeconomics. NBER

Working Paper 21228.

Fuchs-Schündeln, N., and M. Schündeln, 2009. Who stays, who goes, who returns? Economics

of Transition 17, 703-738.

Fulbrook, M., 2005. The people’s state: East German society from Hitler to Honecker. New

Haven: Yale University Press.

Gauly, B., 2017. The intergenerational transmission of attitudes: Analyzing time preferences

and reciprocity. Journal of Family and Economic Issues 38, 293-312.

Gelman, A., and G. Imbens, forthcoming. Why high-order polynomials should not be used in

regression discontinuity designs. Journal of Business and Economic Statistics.

Gill, D., and V. Prowse, 2016. Cognitive ability, character skills, and learning to play equilib-

rium: A level-k analysis. Journal of Political Economy 126, 1619-1676.

Giuliano, P., and A. Spilimbergo, 2014. Growing up in a recession. Review of Economic Studies

81, 787-817.

36



Goldbach, C., and A. Schlüter, 2018. Risk aversion, time preferences, and out-migration.

Experimental evidence from Ghana and Indonesia. Journal of Economic Behavior and Organi-

zation 150, 132-148.

Golsteyn, B.H.H., Grönqvist, H., and L. Lindahl, 2014. Adolescent time preferences predict

lifetime outcomes. Economic Journal 124, F739-F761.

Golsteyn, B., and H. Schildberg-Hörisch, 2017. Challenges in research on preferences and per-

sonality traits: Measurement, stability, and inference. Journal of Economic Psychology 60, 1-6.

Hardardottir, H., 2017. Long term stability of time preferences and the role of the macroeco-

nomic situation. Journal of Economic Psychology 60, 21-36.

Heidemeyer, H., 1994. Flucht und Zuwanderung aus der SBZ/DDR 1945/1949-1961. Düsseldorf,

Droste Verlag.

Heineck, G., and B. Süssmuth, 2013. A different look at Lenin’s legacy: Social capital and risk

taking in the two Germanies. Journal of Comparative Economics 41, 789-803.

Huber, M., 2014. Identifying causal mechanisms (primarily) based on inverse probability

weighting. Journal of Applied Econometrics 29, 920-943.

Hubert, M., 1998. Deutschland im Wandel. Geschichte der deutschen Bevölkerung seit 1815.

Stuttgart, Franz Steiner Verlag.

Humphries, J.E., and F. Kosse, 2017. On the interpretation of non-cognitive skills – What

is being measured and why it matters. Journal of Economic Behavior and Organization 136,

174-185.

Hunt, J., 2006. Staunching emigration from East Germany: Age and the determinants of mi-

gration. Journal of the European Economic Association 4, 1014-1037.

Imai, K., Keele, L., and T. Yamamoto, 2010a. Identification, inference, and sensitivity analysis

for causal mediation effects. Statistical Science 25, 51-71.

Imai, K., Keele, L., and D. Tingley, 2010b. A general approach to causal mediation analysis.

Psychological Methods 15, 309-334.

Imbens, G. W., and K. Kalyanaraman, 2012. Optimal bandwidth choice for the regression

discontinuity estimator. Review of Economic Studies 79, 933-959.

Imbens, G.W., and J.M. Wooldridge, 2009. Recent developments in the econometrics of pro-

gram evaluation. Journal of Economic Literature 47, 5-86.

Knudsen, E., Heckman, J.J., Cameron, J.L., and J.P. Shonkoff, 2006. Economic, neurobiolog-

ical, and behavioral perspectives on building America’s future workforce. PNAS 103, 10155-

10162.

Koch, A., Nafziger, J., and H.S. Nielsen, 2015. Behavioral economics of education. Journal of

Economic Behavior and Organization 115, 3-17.

Kosse, F., Deckers, T., Schildberg-Hörisch, H., and A. Falk, 2016. The formation of prosocial-

37



ity: Causal evidence on the role of the social environment. IZA Discussion Paper No. 9861.

Kosse, F., and F. Pfeiffer, 2013. Quasi-hyperbolic time preferences and their intergenerational

transmission. Applied Economics Letters 20, 983-986.

Krosnick, J.A., and D.F. Alwin, 1989. Aging and susceptibility to attitude change. Journal of

Personality and Social Psychology 57, 416-425.

Laibson, D., 1997. Golden eggs and hyperbolic discounting. Quarterly Journal of Economics

112, 443-478.

Lee, D.S., and D. Card, 2008. Regression discontinuity inference with specification error. Jour-

nal of Econometrics 142, 655-674.

Litz, P., 2007. Der Beitrag des Sports zur Entfaltung der sozialistischen Persönlichkeit. Weis-

sensee Verlag, Berlin.

Marcus, J., 2013. The effect of unemployment on the mental health of spouses: Evidence from

plant closures in Germany. Journal of Health Economics 32, 546-558.

Meier, S., and C.D. Sprenger, 2010. Present-biased preferences and credit card borrowing.

American Economic Journal: Applied Economics 2, 193-210.

Meier, S., and C.D. Sprenger, 2015. Temporal stability of time preferences. Review of Eco-

nomics and Statistics 97, 273-286.

Mischel, W., 2014. The Marshmallow Test: Understanding self-control and how to master it.

London: Transworld Publishers.

Muraven M., 2012. Ego-depletion: theory and evidence. In: Ryan R.M., (Ed.), Oxford Hand-

book of Motivation. Oxford University Press: Oxford.

Necker, S., and A. Voskort, 2014. Politics and parents – Intergenerational transmission of val-

ues after a regime shift. European Journal of Political Economy 36, 177-194.

Nguyen, Q., 2011. Does nurture matter: Theory and experimental investigation on the effect

of working environment on risk and time preferences. Journal of Risk and Uncertainty 43,

245-270.

Nunn, N., and L. Wantchekon, 2011. The slave trade and the origins of mistrust in Africa.

American Economic Review 101, 3221-3252.

Ockenfels, A., and J. Weimann, 1999. Types and patterns: an experimental EastWest compar-

ison of cooperation and solidarity. Journal of Public Economics 71, 275-287.

O’Donoghue, T., and M. Rabin, 1999. Doing it now or later. American Economic Review 89,

106-124.

O’Donoghue, T., and M. Rabin, 2006. Optimal sin taxes. Journal of Public Economics 90,

1825-1849.

O’Donoghue, T., and M. Rabin, 2015. Present bias: Lessons learned and to be learned. Amer-

ican Economic Review Papers & Proceedings 105, 273-279.

38



Oster, E., forthcoming. Unobservable selection and coefficient stability: Theory and evidence.

Journal of Business and Economic Statistics.

Patton, J.H., Stanford, M.S., and E.S. Barratt, 1995. Factor structure of the Barratt impul-

siveness scale. Journal of Clinical Psychology 51, 768-774.

Pinger, P.R., 2017. Predicting experimental choice behavior and life outcomes from a survey

measure of present bias. Economics Bulletin, Volume 37(3), 2162-2172.

Rainer, H., and T. Siedler, 2009. Does democracy foster trust? Journal of Comparative Eco-

nomics 37, 251-269.

Redding, S.J., and D.M. Sturm, 2008. The costs of remoteness: Evidence from German division

and reunification. American Economic Review 98, 1766-1797.

Richter, D., and J. Schupp, 2014. SOEP - TIMEPREF: Dataset on the Economic Behavior

Experiment on Time Preferences in the 2006 SOEP Survey. SOEP Survey Papers 224.

Ritschl, A., and M. Spoerer, 1997. Das Bruttosozialprodukt in Deutschland nach den amtlichen

Volkseinkommens- und Sozialproduktstatistiken 1901-1995. GESIS Datenarchiv Köln, Studi-

ennummer 8137.

Robson, A.J., and L. Samuelson, 2011. The evolutionary foundations of preferences. In: Ben-

habib, J., Bisin, A., Jackson, M. (Eds.). Handbook of Social Economics 1A, 221-310.

Saunders, A., 2007. Honecker’s children: Youth and patriotism in East(ern) Germany, 1979-

2002. Manchester: Manchester University Press.

Schäfgen, K., 1998. Die Verdoppelung der Ungleichheit. Sozialstruktur und Geschlechter-

verhältnisse in der Bundesrepublik und in der DDR. Humboldt-University Berlin, Germany.

Schildberg-Hörisch, H., 2018. Are risk preferences stable? Journal of Economic Perspectives

32, 135-154.

Sleifer, J., 2006. Berechnungen zum Bruttoinlandprodukt der DDR. GESIS Datenarchiv Köln,

Studiennummer 8466.

Sozou, P.D., 1998. On hyperbolic discounting and uncertain hazard rates. Proceedings of the

Royal Society B 265, 2015-2020.

Specht, J., Luhmann, M., and C. Geiser, 2014. On the consistency of personality types across

adulthood: latent profile analyses in two large-scale panel studies. Journal of Personality and

Social Psychology 107, 540-556.

Stigler, G.J., and G.S. Becker, 1977. De gustibus non est disputandum. American Economic

Review 67, 76-90.

Sutter, M., Kocher, M.G., Glätzle-Rützler, D., and S.T. Trautmann, 2013. Impatience and

uncertainty: Experimental decisions predict adolescent’s field behavior. American Economic

Review 103, 510-531.

Sutter, M., Yilmaz, L., and M. Oberauer, 2015. Delay of gratification and the role of defaults

39



– An experiment with kindergarten children. Economics Letters 137, 21-24.

Tabellini, G., 2008. The scope of cooperation: Values and incentives. Quarterly Journal of

Economics 123, 905-950.

Tangney, J.P., Baumeister, R.F., and A.L. Boone, 2004. High self-control predicts good ad-

justment, less pathology, better grades, and interpersonal success. Journal of Personality 72,

271-322.

Vischer, T., Dohmen, T., Falk, A., Huffman, D., Schupp, J., Sunde, U., and G.G. Wagner,

2013. Validating an ultra-short survey measure of patience. Economics Letters 120, 142-145.

Voigtländer, N., and H.J. Voth, 2012. Persecution perpetuated: The medieval origins of anti-

semitic violence in Nazi Germany. Quarterly Journal of Economics 127, 1339-1392.

Von Ehrlich, M., and T. Seidel, forthcoming. The persistent effects of place-based policy: Evi-

dence from the West-German Zonenrandgebiet. American Economic Journal: Economic Policy.

Wang, M., Rieger, M.O., and T. Hens, 2016. How time preferences differ: Evidence from 53

countries. Journal of Economic Psychology 52, 115-135.

Wolf, N., 2009. Was Germany ever united? Evidence from intra- and international trade,

1885-1933. Journal of Economic History 69, 846-881.

Zumbuehl, M., Dohmen, T., and G. Pfann, 2013. Parental investment and the intergenerational

transmission of economic preferences and attitudes.SOEPpapers 570-2013.

40



Appendix

A Exploring the relationship of the SOEP proxy for

present bias (impulsivity) and an experimental mea-

sure of present bias

In 2006, experiments were conducted with a subsample of the SOEP to elicit their time prefer-

ences. In the experiment that we rely upon for our experimental measure of present bias, 526

individuals were asked to indicate their 20 choices from 20 sets of two alternatives, where the

alternatives differed in their level of payout and their payout period. The difference in the pay-

out period was fixed at one month. In a first task, participants were presented with the option

of either receiving 200 euros immediately or some payment x > 200 euros in one month. In a

second task, the decision was between 200 euros in 12 months and some payment x > 200 euros

in 13 months. The level of x took on 20 values that were presented in an ascending order and

the subjects were asked to state their choice for each of the 20 sets of two alternatives (Richter

and Schupp 2014). After the experiment, a randomly selected subgroup of participants (11%)

received 200 euros or a random x > 200 euros depending on their choice in the experiment.

The set of alternatives at which individuals switch from preferring the sooner payment to

preferring the later payment gives us their switching point and is interpreted as a point of

indifference. We observe valid switching points for 374 participants aged 18 to 80 years. These

switching points allow us to calculate an experimental measure of present bias. For illustration,

suppose that an individual is indifferent between the 200 euros immediately and x euros in one

month (i.e., that u(200) = βδu(x), where we use the notation from Footnote 2) and that this

individual is indifferent between 200 euros in 12 months and y euros in 13 months (i.e., that

u(200) = δu(y)). For this individual, we can infer β = u(y)/u(x).

In a next step, we use an ordinary least squares specification to estimate the correlation

between the experimental measure of present bias and our survey proxy for present bias (i.e., the

SOEP measure for impulsivity) using age, age squared, and gender as additional covariates. For

a subset of 322 of the 374 experiment participants with valid switching points and information

about the place of residence in 1989, the SOEP survey in 2008 contains information on our proxy

for present bias. We find a significant correlation between our survey measure for present bias

and the experimental measure of present bias. The correlation remains significant and nearly

unchanged when we add age, age squared, gender, and survey information on patience and risk

(see Table 11 for complete results). Tobit and robust regression exercises yield similar results.

Considering only subjects that were FRG residents in 1989, we find that the results from the

41



pooled sample re-emerge. Considering only subjects that were GDR residents in 1989, we find

that the sign and the size of the coefficient remain intact. However, these coefficients are not

statistically significant, which may be related to the very small number of observations.

42



(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

P
o
o
le

d
P

o
o
le

d
P

o
o
le

d
F

R
G

F
R

G
F

R
G

G
D

R
G

D
R

G
D

R

sa
m

p
le

sa
m

p
le

sa
m

p
le

sa
m

p
le

sa
m

p
le

sa
m

p
le

sa
m

p
le

sa
m

p
le

sa
m

p
le

Im
p
u
ls

iv
it

y
-0

.0
0
1
∗

-0
.0

0
2
∗

-0
.0

0
1

+
-0

.0
0
2
∗

-0
.0

0
2
∗

-0
.0

0
1

-0
.0

0
1

-0
.0

0
1

-0
.0

0
1

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

A
g
e

-0
.0

0
1

-0
.0

0
1

-0
.0

0
1

-0
.0

0
1

-0
.0

0
0

0
.0

0
0

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

A
g
e2

0
.0

0
0

0
.0

0
0

0
.0

0
0

0
.0

0
0

-0
.0

0
0

-0
.0

0
0

(0
.0

0
0
)

(0
.0

0
0
)

(0
.0

0
0
)

(0
.0

0
0
)

(0
.0

0
0
)

(0
.0

0
0
)

M
a
le

0
.0

0
1

0
.0

0
2

0
.0

0
1

0
.0

0
2

0
.0

0
2

0
.0

0
3

(0
.0

0
3
)

(0
.0

0
3
)

(0
.0

0
3
)

(0
.0

0
3
)

(0
.0

0
7
)

(0
.0

0
7
)

P
a
ti

en
ce

-0
.0

0
1

-0
.0

0
1

-0
.0

0
1

-0
.0

0
1

-0
.0

0
1

-0
.0

0
1

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
1
)

R
is

k
-0

.0
0
1

-0
.0

0
1

0
.0

0
1

(0
.0

0
1
)

(0
.0

0
1
)

(0
.0

0
2
)

N
3
2
2

3
2
2

3
2
2

2
7
7

2
7
7

2
7
7

4
5

4
5

4
5

R
2

0
.0

1
6

0
.0

2
4

0
.0

3
0

0
.0

1
7

0
.0

2
5

0
.0

3
2

0
.0

0
9

0
.0

4
5

0
.0

5
2

N
o
te
s
:

W
e

u
se

ex
p

er
im

en
ta

l
d

a
ta

fr
o
m

2
0
0
6

a
n

d
S

O
E

P
su

rv
ey

d
a
ta

fr
o
m

2
0
0
8
.

P
a
ra

m
et

er
es

ti
m

a
te

s
co

m
e

fr
o
m

o
rd

in
a
ry

le
a
st

sq
u

a
re

s
re

g
re

ss
io

n
s.

T
h

e
d

ep
en

d
en

t
v
a
ri

a
b

le
is

th
e

ex
p

er
im

en
ta

l
m

ea
su

re
fo

r
p
re
se
n
t
b
ia
s.

R
o
b

u
st

st
a
n
d

a
rd

er
ro

rs

a
re

d
o
cu

m
en

te
d

;
+
p
<

0
.1

0
,
∗
p
<

0
.0

5
,
∗∗

p
<

0
.0

1
.

T
a
b
le

1
1
:

C
o
rr

el
a
ti

o
n

o
f

Im
p
u
ls

iv
it

y
a
n
d

E
x
p

er
im

en
ta

l
M

ea
su

re
o
f

P
re

se
n
t

B
ia

s

43



B Robustness Checks

Present bias
GDR -0.178∗∗

(0.060)
Moved West during Separation 0.060

(0.211)
Moved West -0.147

(0.170)
Moved East -0.117

(0.334)
Age 0.020

(0.016)

Age2 -0.000∗

(0.000)
Male -0.302∗∗

(0.054)
Raised in a large city 0.237∗∗

(0.077)

Raised in a medium city 0.157+

(0.083)
Raised in a small city -0.033

(0.075)
Mother: Secondary lower school 0.183

(0.144)
Mother: Secondary intermediate school 0.157

(0.164)
Mother: Secondary higher school -0.192

(0.242)
Mother: No vocational degree -0.072

(0.101)
Mother: Vocational degree -0.081

(0.103)
Mother: Technical school 0.116

(0.178)
Mother: University degree 0.151

(0.310)
Father: Secondary lower school -0.332∗

(0.136)

Father: Secondary intermediate school -0.266+

(0.161)
Father: Secondary higher school -0.156

(0.194)
Father: No vocational degree 0.256∗

(0.127)
Father: Vocational degree 0.214∗

(0.105)
Father: Technical school 0.224

(0.161)
Father: University degree -0.041

(0.201)
N 8,341

R2 0.018

Notes: We use SOEP data from 2008 and 2013 for respondents who also
participated in one of the survey years 1990 to 1993. Parameter estimates
are from ordinary least squares specifications. The dependent variable is
the SOEP present bias measure. The specification includes all covariates
from Table 3. Robust standard errors are clustered at the individual level
and reported in parentheses; + p < 0.10, ∗ p < 0.05, ∗∗ p < 0.01.

Table 12: Taking Account of East-West Migration before 1989: Ordinary Least Square Regres-
sions

44



Present bias
GDR -0.771

(0.230)∗∗

[0.204]∗∗

N 2,621
R2 0.020

Notes: We use SOEP data from 2008 and 2013 for respondents who
also participated in one of the survey years 1990 to 1993. Parame-
ter estimates are based on Specification 1 detailed in Section 3.2.2.
The dependent variable is the SOEP present bias measure. Ro-
bust standard errors clustered at the individual (county) level are
reported in parentheses (square brackets); + p < 0.10, ∗ p < 0.05,
∗∗ p < 0.01.

Table 13: Taking Account of East-West Migration before 1989: GRDD Regressions

Present bias
GDR × Brandenburg -0.446∗∗

(.100)
GDR × Meck. West. Pom. -0.323∗

(.126)
GDR × Saxony -0.201∗∗

(.071)
GDR × Saxony-Anhalt -0.381∗∗

(.096)
GDR × Thuringia -0.125

(.090)
GDR × Berlin -0.350∗

(.158)
GDR × state info miss -0.149∗∗

.035
N 33,109
R2 0.0163

Notes: We use SOEP data from 2008 and 2013. Parameter esti-
mates are from ordinary least squares specifications. The depen-
dent variable is the SOEP present bias measure. The specifications
include all covariates from Table 3. Robust standard errors are
clustered at the individual level and presented in parentheses; +

p < 0.10, ∗ p < 0.05, ∗∗ p < 0.01.

Table 14: Present Bias and GDR Treatment: GDR-states Interactions

45


	Introduction
	Data
	Research Design
	The Formation of Time Preferences and Potential Regime Influence
	Identification Strategy: The German Separation and Reunification as a Natural Experiment
	Ordinary Least Squares Regressions and Test of Selection on Unobservables
	Geographic Regression Discontinuity Design


	Empirical Results
	Results from Ordinary Least Squares Regressions
	Results from Geographic Regression Discontinuity Design
	Exploring Variation in Treatment Intensity: Duration of Regime Exposure
	Robustness Checks

	Exploring Implications of GDR Experience
	Conclusion
	Exploring the relationship of the SOEP proxy for present bias (impulsivity) and an experimental measure of present bias
	Robustness Checks