General Education

Mmr00

ContentServer4.pdf

Home >Business & Finance homework help >Marketing homework help >General Education

Direct and indirect effects of weather experiences on life satisfaction � which role for climate change expectations?

Daniel Osberghaus a * and Jan K€uhlingb

a Centre for European Economic Research (ZEW), Mannheim, Germany;

b Department of

Economics, Carl von Ossietzky University of Oldenburg, Oldenburg, Germany

(Received 19 March 2015; final version received 31 December 2015)

This paper deals with the effect of (1) damage experience from extreme weather events and (2) expectations concerning future climate change on subjective well-being (SWB). We use data from a large representative survey carried out amongst German households. The effect of experienced weather events on the SWB of the heads of households is significant only in the case of heat waves; the same cannot be said for storms, heavy rain, and floods. Concerns about future climate change in households have a substantial negative impact on current SWB. In addition, we divide the impact of experience into direct and indirect effects of damage, deduced from the impact of experience on expectations regarding future climate change. Both direct and indirect effects of weather experiences are quantified. It becomes apparent that the indirect effect is significant, but small when compared to the direct effect.

Keywords: climate change; subjective well-being; extreme weather events; household survey

1. Introduction and literature

Climate change and extreme weather events (EWEs) such as heat waves, storms, and

floods affect the living conditions of households and individuals worldwide.

Climatologists expect to see an average global temperature rise of between 1.0 and

3.7 �C, relative to 1986�2005 (IPCC 2014), by 2100. This is likely to result in an increase in the frequency and severity of EWEs (Field et al. 2012). This paper concerns

the effect of EWEs and climate change on the subjective well-being (SWB) of

individuals. We analyse the SWB effects of (1) weather-related material and damage to

health experiences and (2) expectations about future climate change. While the experience

analysis focuses on the impact of past events on current SWB, the expectations of climatic

conditions take account of individuals’ current concerns regarding the future effects of

global warming.

While there is a large body of literature which addresses the SWB effects of weather

events, we are not aware of any study which has assessed the impact of expectations

regarding future climate change, on current SWB. It is conceivable, however, that

relatively high levels of concern (i.e. expectation that climate change will have negative

impacts) correlate with a significant downward shift in the current SWB. We will test and

quantify this hypothesis.

*Corresponding author. Email: osberghaus@zew.de

� 2016 University of Newcastle upon Tyne

Journal of Environmental Planning and Management, 2016

Vol. 59, No. 12, 2198�2230, http://dx.doi.org/10.1080/09640568.2016.1139490

mailto:osberghaus@zew.de

http://dx.doi.org/10.1080/09640568.2016.1139490

A further relevant question concerns the interaction between climate change

experience and expectations. Existing literature on climate change risk perception

provides empirical evidence suggesting that experience of EWEs may lead to a higher

level of concern about future global warming (Akerlof et al. 2013; Bichard and

Kazmierczak 2012; Spence et al. 2011; van der Linden 2015). Note, however, that

Whitmarsh 2008 does not report such an effect. Assuming the existence of this relation,

the effect of damage experience on SWB may be separated into a direct, and an indirect

effect. The former refers to the immediate effect of a negative event, whilst the latter, the

secondary effect of an experience on concerns about future climate change. Hence, the

introduction of concern about future climate change into happiness research allows

deeper analysis of the interaction between damage experience, concern about future

outcomes, and SWB.

Figure 1 depicts the hypothesised relations between damage experience, climate

change expectations, and SWB.

There is a growing body of literature concerning the effect of experienced climate

change and EWEs on SWB. Research in this field has yielded heterogeneous regression

analyses which differ not only in terms of their geographical and time dimensions, but

also in terms of their methodological frameworks. 1 Furthermore, these studies use

various climate and EWE variables as determinants of SWB, which leads to the

identification of different well-being effects (see Table 1).

While some of these studies focus on single countries and identify country-specific

effects (e.g. Frijters and van Praag 1998), other papers adopt a more global perspective,

analysing groups of countries (e.g. Maddison and Rehdanz 2011). The majority of such

studies support the hypothesis according to which EWEs and climate change have a

negative effect on SWB. Having said this, in certain countries, SWB may in fact improve

in view of predicted changes in climate. A number of studies, such as Kimball et al.

(2006) or Kountouris and Remoundou (2011), indicate that adaptation to changed EWE

patterns is possible and that this may mitigate effects on SWB. Although the features

considered, and the methodologies used in these studies vary, they have in common that

they fail to acknowledge EWE expectations.

Our contribution to the literature is, therefore, threefold. First, we examine the relation

of SWB and past EWEs in Germany (thus far, the SWB analyses for Germany have

Figure 1. Hypothesised effects between past damage experience, expectations of future climate change, and subjective well-being. Direct relationships are depicted by solid arrows, indirect ones by dashed arrows.

Journal of Environmental Planning and Management 2199

T a b le 1 .

L it e ra tu re

o v e rv ie w : E W E e ff e c ts o n S W B .

P a p e r

D a ta

L o c a ti o n

T im

e C li m a te a n d / o r E W E

v a ri a b le (s )

C li m a te a n d / o r E W E e ff e c ts o n S W B

F ri jt e rs a n d v a n

P ra a g (1 9 9 8 )

C ro ss -s e c ti o n

a n d ti m e

se ri e s

R u ss ia

1 9 9 3 �1

9 9 4

T e m p e ra tu re , h u m id it y ,

p re c ip it a ti o n , su n sh in e

-N e g a ti v e e ff e c ts o f h a rs h w in te rs a n d h ig h h u m id it y

in c o m b in a ti o n w it h h ig h te m p e ra tu re s

-P o si ti v e e ff e c ts fr o m

m o re

su n h o u rs a n d h ig h e r

te m p e ra tu re s

K im

b a ll e t a l.

(2 0 0 6 )

C ro ss -s e c ti o n

a n d ti m e

se ri e s (P a n e l)

U S A

A u g u st 2 0 0 5

u n ti l O c to b e r

2 0 0 5

H u rr ic a n e “ K a tr in a ”

-N e g a ti v e im

p a c ts o f h u rr ic a n e s;

-P ro n o u n c e d e ff e c ts in

a ff e c te d a re a s,

-A d a p ta ti o n e ff e c ts im

m e d ia te ly

a ft e r h u rr ic a n e

B re re to n ,

C li n c h , a n d

F e rr e ir a

(2 0 0 8 )

D is a g g re g a te d

c ro ss -s e c ti o n

Ir e la n d

2 0 0 1

W in d sp e e d , M in -t e m p e ra tu re

in Ja n u a ry , M a x -t e m p e ra tu re

in Ju ly

-N e g a ti v e e ff e c ts fr o m

h ig h e r w in d sp e e d

-P o si ti v e e ff e c ts fr o m

in c re a se s b o th

in M in -

te m p e ra tu re

in Ja n u a ry

a n d M a x -t e m p e ra tu re

in Ju ly

F e rr e ir a a n d

M o ro

(2 0 1 0 )

D is a g g re g a te d

c ro ss -s e c ti o n

Ir e la n d

2 0 0 1

R a in fa ll , M in -t e m p e ra tu re

in Ja n u a ry , M a x -t e m p e ra tu re

in Ju ly

-P o si ti v e e ff e c ts fr o m

in c re a se s b o th

in M in -

te m p e ra tu re

in Ja n u a ry

a n d M a x -t e m p e ra tu re

in Ju ly

M a c K e rr o n a n d

M o u ra to

(2 0 1 3 )

D is a g g re g a te d

c ro ss -s e c ti o n

(P a n e l)

U K

A u g u st 2 0 1 0

u n ti l

F e b ru a ry

2 0 1 1

D a y li g h t, sn o w , su n sh in e , fo g ,

ra in

-P o si ti v e e ff e c ts o f su n sh in e a n d e x p o si ti o n to

g re e n

o r n a tu ra l e n v ir o n m e n ts

-N e g a ti v e e ff e c ts o f fo g a n d ra in

C a rr o ll , F ri jt e rs ,

a n d S h ie ld s

(2 0 0 9 )

C ro ss -s e c ti o n

a n d ti m e

se ri e s

A u st ra li a

2 0 0 1 �2

0 0 4

S e v e re

sp ri n g d ro u g h t

-N e g a ti v e e ff e c ts fo r in d iv id u a ls li v in g in

ru ra l a re a s

A m b re y a n d

F le m in g

(2 0 1 1 )

D is a g g re g a te d

c ro ss -s e c ti o n

A u st ra li a

2 0 0 5

M e a n ra in fa ll , M in - a n d M a x -

te m p e ra tu re , w in d sp e e d ,

su n sh in e

-P o si ti v e e ff e c ts o f in c re a se d se a so n a l v a ri a ti o n in

c li m a te v a ri a b le s

-N e g a ti v e e ff e c ts o f su n sh in e

F e d d e rs e n ,

M e tc a lf e , a n d

W o o d e n

(2 0 1 2 )

C ro ss -s e c ti o n

a n d ti m e

se ri e s (P a n e l)

A u st ra li a

2 0 0 1 �2

0 1 0

S o la r e x p o su re , m e a n a n d

e x tr e m e te m p e ra tu re , ra in fa ll ,

w in d sp e e d a n d d ir e c ti o n

-N o e ff e c t o f c li m a te v a ri a b le s a s o p p o se d to

w e a th e r

v a ri a b le s

C u ~ n a d o a n d d e

G ra c ia

(2 0 1 3 )

D is a g g re g a te d

c ro ss -s e c ti o n

S p a in

2 0 0 8

Ja n u a ry

a n d Ju ly

M in - a n d M a x -

te m p e ra tu re , Ja n u a ry

a n d Ju ly

p re c ip it a ti o n

-N e g a ti v e im

p a c ts o f h ig h e r M in -t e m p e ra tu re

a n d

m o re

p re c ip it a ti o n in

Ju ly

(c o n ti n u e d )

2200 D. Osberghaus and J. K€uhling

T a b le 1 .

(C o n ti n u e d )

P a p e r

D a ta

L o c a ti o n

T im

e C li m a te a n d / o r E W E

v a ri a b le (s )

C li m a te a n d / o r E W E e ff e c ts o n S W B

K o u n to u ri s a n d

R e m o u n d o u

(2 0 1 1 )

C ro ss -s e c ti o n

a n d ti m e

se ri e s

S p a in , P o rt u g a l,

It a ly , M e d i-

te rr a n e a n

F ra n c e

1 9 8 6 �1

9 9 3 a n d

2 0 0 1 �2

0 0 3

F o re st fi re

in c id e n ts

-N e g a ti v e im

p a c ts o f fo re st fi re

in c id e n t c a u se d b y

d ro u g h t

-A d a p ta ti o n e ff e c ts a ft e r fo re st fi re

in c id e n t

M u rr a y ,

M a d d is o n ,

a n d R e h d a n z

(2 0 1 3 )

D is a g g re g a te d

c ro ss -s e c ti o n

1 9 E u ro p e a n

c o u n tr ie s

1 9 9 9 �2

0 0 0

R e la ti v e h u m id it y , su n sh in e

h o u rs , m e a n te m p e ra tu re ,

w in d sp e e d , ra in

a n d fr o st

d a y s, p re c ip it a ti o n

-P o si ti v e e ff e c ts o f su n sh in e a n d lo w e r re la ti v e

h u m id it y in

a re a s w it h lo w e r le v e ls o f su n sh in e a n d

h ig h e r le v e ls o f re la ti v e h u m id it y

-H ig h e r (l o w e r) w e lf a re

le v e ls fo r M e d it e rr a n e a n

(S c a n d in a v ia n ) c li m a te c o n d it io n s

L u e c h in g e r a n d

R a sc h k y

(2 0 0 9 )

C ro ss -s e c ti o n

a n d ti m e

se ri e s

1 6 E u ro p e a n

c o u n tr ie s

1 9 7 3 �1

9 9 8

F lo o d in g e v e n ts

-N e g a ti v e e ff e c ts o f fl o o d in g e v e n ts

-M it ig a ti n g e ff e c ts o f

m a n d a to ry

in su ra n c e

G r€ u n a n d

G ru n e w a ld

(2 0 1 0 )

C ro ss -s e c ti o n

a n d ti m e

se ri e s (P a n e l)

1 8 L a ti n

A m e ri c a n

c o u n tr ie s

1 9 9 5 �2

0 0 8

M e a n a n d e x tr e m e te m p e ra tu re ,

p re c ip it a ti o n , c lo u d c o v e re d

d a y s, v a p o u r p re ss u re

-P o si ti v e e ff e c ts o f h ig h e r te m p e ra tu re s in

th e c o ld e st

m o n th s

-N e g a ti v e e ff e c ts o f c lo u d c o v e re d d a y s

R e h d a n z a n d

M a d d is o n

(2 0 0 5 )

C ro ss -s e c ti o n

d a ta w it h

ti m e

v a ri a ti o n

6 7 c o u n tr ie s

w o rl d w id e

1 9 8 4 �2

0 0 1

M e a n p re c ip it a ti o n a n d

te m p e ra tu re , m e a n

te m p e ra tu re

o f c o ld e st a n d

h o tt e st m o n th s, m e a n

p re c ip it a ti o n in

d ri e st a n d

w e tt e st m o n th s, n u m b e r o f

c o ld , h o t, d ry , a n d w e t m o n th s

-P o si ti v e e ff e c ts o f lo w e r te m p e ra tu re s in

su m m e r,

h ig h e r te m p e ra tu re s in

w in te r, a n d h ig h e r

p re c ip it a ti o n in

th e d ri e st m o n th s

-N e g a ti v e e ff e c ts fo r a m a jo ri ty

o f c o u n tr ie s

su ff e ri n g fr o m

e x p e c te d c li m a te c h a n g e

M a d d is o n a n d

R e h d a n z

(2 0 1 1 )

C ro ss -s e c ti o n

d a ta w it h

ti m e

v a ri a ti o n

7 9 c o u n tr ie s

w o rl d w id e

1 9 8 1 �2

0 0 8

n u m b e r o f m o n th s w it h a v e ra g e

m e a n te m p e ra tu re

a b o v e /

b e lo w 1 8 .3

� C

-N e g a ti v e e ff e c ts fo r d e v ia ti o n s fr o m

a b a se

te m p e ra tu re

o f 1 8 .3

� C in

b o th

d ir e c ti o n s

-h ig h e st w e lf a re

lo ss e s fo r A fr ic a n c o u n tr ie s th ro u g h

g lo b a l w a rm

in g

B e c c h e tt i,

C a st ri o ta , a n d

A n d r� e s

(2 0 0 7 )

C ro ss -s e c ti o n

d a ta w it h

ti m e

v a ri a ti o n

C it ie s in

8 0

c o u n tr ie s

w o rl d w id e

1 9 9 5 �2

0 0 4

M e a n te m p e ra tu re , n u m b e r o f

m o n th s w it h te m p e ra tu re

a b o v e 2 0 � C

/ b e lo w 0 � C

, w in d

sp e e d , n u m b e r o f fo g g y a n d

ra in y d a y s

-N e g a ti v e e ff e c ts o f h ig h e r w in d sp e e d , h ig h e r

te m p e ra tu re s, a n d m o re

fo g g y d a y s

-I n v e rt e d U -s h a p e d re la ti o n sh ip

b e tw e e n ra in y d a y s

a n d S W B w it h a tu rn in g p o in t a ro u n d 2 2 0 a n n u a l

ra in y d a y s

Journal of Environmental Planning and Management 2201

largely concentrated on the impact of long-term climate variables). Second, we quantify

the relation between concern about future climate change and the current SWB. To our

knowledge, this study is the first attempt in this regard. Finally, we separate the SWB

effect of damage experience into a direct effect, and an indirect effect, deduced from the

impact of experience on expectations regarding future climate change.

2. Theoretical model

A key finding of literature on happiness research is that data on SWB may be used as an

empirical approximation of utility (see e.g. Frey and Stutzer 2002). In view of this, it is

possible to incorporate the interrelationship of SWB, damage experiences, and damage

expectations into a corresponding utilitarian framework, which enables differentiation

between the direct and indirect effects of EWEs, and interpretation of these in terms of

their utility consequences.

In the first instance, damage experiences may have direct utility consequences by

means of health-related effects such as cardiovascular complaints during heat waves, for

example, and economic effects such as financial and material losses as a result of storms,

flooding, or heavy rain. One can assume that the direct utility consequences depend on

the type, severity, and duration of the damage experience itself.

In the second instance, indirect utility consequences may result from damage

experiences where these impact on damage expectations. A past experience of an EWE,

for example, may alter perceptions of future climatic conditions and thus lead to fears of

detrimental health effects, global productivity reductions, and personal income losses.

The above considerations are reflected in the following theoretical framework:

U D f ðV; ZÞ (1a) Z D gðVÞ (1b)

Whilst U denotes present utility, V stands for past damage experience, and Z for damage

expectations in the future. The theoretical model given in Equations (1a) and (1b) is the

mathematical analogue to Figure 1 and provides the theoretical basis for separating and

estimating the effects EWEs have on individual SWB:

dV D @U

@V C @U

@Z � @Z @V

(2)

In Equation (2), the left-hand side measures the total effect of EWE. The first summand

on the right-hand side is the direct effect of EWE experience, whereas the second

summand is its indirect effect, whereby experience results in a change in expectations

regarding negative climate impacts. It should be noted that the direct effect is the

marginal utility of experience, whereas the indirect effect consists of the marginal utility

of expectation multiplied by the marginal effect that experience has on expectations.

In view of this, the utilitarian framework separates direct and indirect effects and

indicates their components. The model also reflects our assumption regarding causality.

We believe that, in theory, it is more likely that experiences and expectations affect LS,

rather than the other way around. In addition, the utilitarian framework offers the

opportunity for monetisation of the direct and indirect effects. This is done by combining

2202 D. Osberghaus and J. K€uhling

the marginal utilities of EWE experiences and expectations, with the marginal utility of

income. The utilitarian framework then allows the marginal rate of substitution between

experience/expectations and income or infra-marginal measures (equivalent/

compensating variation), to be calculated as monetary values (see e.g. Welsch and

K€uhling 2009).2

3. Data

We use cross-section data taken from a survey carried out amongst German households.

In total, out of a representative sample of approximately 10,000 households, 6,404

households responded and were interviewed via either an online or a TV-based

questionnaire (Osbergaus 2015). Given that only heads of households were interviewed,

the sample is largely representative in terms of households, but not on the level of

individuals. 3 The survey was conducted in October and November 2012. Towards the

end of the survey period, the hurricane Sandy made landfall on the US East Coast.

This event, and the resulting substantial damages, received a great deal of coverage in the

German media. After the hurricane Sandy made landfall, 4.4% of the sample was

interviewed. As a cross section, the dataset cannot be said to directly reflect this time

dimension. Having said this, the key variables (SWB, experience, expectations) are

quasi-temporal, as participants were explicitly asked for current SWB, damage events

experienced in the past, and expectations for the future. An aggregated overview of the

data and more information on the survey, including the questionnaire (in the German

language), are available in Osberghaus, Schwirplies, and Ziegler 2013. For the present

analysis, we use the key variables presented in Table 2 and a number of control variables

presented in Table 1.1. Due to the limited availability of data, we control for health by

using dummy variables for normal weight, overweight, etc., and the body mass index

(BMI) of respondents. Data from the German socio-economic panel shows that

Table 2. Descriptive statistics of key variables.

Variable in the model Variable in the data

Mean / in case of binary variables: share of “yes”-responses Std. dev. Min. Max. Obs.

Subjective well- being (present)

Self-rated life satisfaction (LS)

7.177 1.960 0 (totally dissatisfied)

10 (totally satisfied)

6,397

Damage experience (past)

Participant has already experienced financial or health damage by the following:

Heat wave 0.040 0.195 0 (no) 1 (yes) 6,366

Storm 0.226 0.418 0 (no) 1 (yes) 6,367

Heavy rain 0.267 0.442 0 (no) 1 (yes) 6,365

Flood 0.115 0.319 0 (no) 1 (yes) 6,368

Climate change expectation (future)

Expected consequences of climate change on individual living conditions in the next decades

3.550 0.660 1 (very positive)

5 (very negative)

5,336

Journal of Environmental Planning and Management 2203

conventional health variables, such as self-reported status of general health, have a high

and significant correlation with the BMI variables. Table 1.2 contains a correlation

matrix of the key variables. 4 The key variables are described in more detail later.

3.1. Subjective well-being

SWB is measured by a single question, the first item of the questionnaire, in order that

answers are not affected by other questions and are not subject to an order bias. Participants

were asked to rate their current individual life satisfaction (LS) on an 11-point Likert scale

ranging from “totally dissatisfied” to “totally satisfied”. 5 This approach was deemed as a

valid and efficient method to obtain SWB, i.e. by Diener et al. (1985). 6 The distribution is

left skewed, a typical pattern for this kind of formulation. While other measures exist, i.e. the

use of multiple items instead of one single question, the use of this single item for eliciting

LS is broadly accepted in relevant literature (see literature cited in Section 1). 7

3.2. Damage experience

In order to measure the past damage experience, the participants were asked to state whether

they had suffered any financial or health damage as a result of heat waves, storms, heavy

rain, or flooding. The terms, “heat wave”, “storm”, and “heavy rain” were further explained

by the means of short illustrative situations indicating how the event may affect personal life

(see Table 1.5). The health damage was restricted to cases where participants consulted a

doctor. The data is, therefore, an objective measure of the stated damage occurring due to

weather events, albeit without indicating the nature, severity, or time of the damage. This is

due to the fact that the questionnaire was to be kept short and simple. 8

3.3. Climate change expectation

In order to measure expectations regarding future climate change, participants were asked

to rate the expected consequences of climate change on their personal living conditions in

the next decades on a five-point Likert scale ranging from “very negative” to “very

positive”. This approach takes a broad perspective in terms of what exactly constitutes

climate change (i.e. without a focus on any specific impact), but puts the focus on

personal conditions. 9

4. Empirical strategy

The empirical analogue to Figure 1 and the theoretical model (Equations (1a) and (1b))

can be formulated as below. Looking first at Equation (1a), we assume that data on LS is

a proxy for utility. The data on damage experience is a binary variable, and data on

damage expectations is an ordinal variable with five values. The empirical analogue to

Equation (1a) is, therefore:

LSi D a C b�Ei C X5 j D 1

gj�Dij C d�Xi C ei Dij D 1 if Ci D j otherwise Dij D 0

(3)

Ei is a dummy variable taking the value of 1 if observation i has experienced any climate-

related damage. Dij is a set of five dummy variables which take the value of 1 if

2204 D. Osberghaus and J. K€uhling

observation i exhibits damage expectations (Ci ) of the level j , where j D 1; :::; 5. The parameters a, b, gj, and d are coefficients which can be estimated. To avoid perfect

multicollinearity, one of the gj must be set at zero (in the following, g3 D 0). Xi denotes a set of control variables and ei the error term. Two testable hypotheses can, therefore, be formulated with respect to Equation (3):

Hypothesis (a): Individuals who have suffered financial or health damage from

EWE in the past tend to exhibit lower LS today (b < 0).

Hypothesis (b): Individuals who expect negative impacts of climate change in the

future tend to exhibit lower LS today (g4;5 < 0, as g3 D 0 denotes neutral climate expectations).

In order to test hypotheses (a) and (b), we use ordinary least squares (OLS)

regressions and ordered probit regressions as robustness checks. 10

As can be seen from

Table 1.1, a part of the variable set Xi is personal attitude variables which capture the

subjective importance of certain topics for the respondent, including environmental

issues and the individual’s economic situation. Such attitude variables have proved to be

important determinants of LS and have significance beyond those correlated with

inherited personality traits (Ferrer-i-Carbonell and Gowdy 2007). These attitude variables

are, therefore, included in the regression analyses for practical reasons; they may work as

proxies for otherwise-unobserved personality trait variables (see e.g. Welsch and K€uhling 2010).

Despite the inclusion of a number of personality traits as control variables, there is a

risk of endogeneity in the estimated relationships. There are at least three potential

reasons for endogeneity in our regressions. First, endogeneity may be caused by

unobserved confounding variables which correlate with both the dependent variable and

the regressor. Despite the inclusion of control variables, we cannot, therefore, rule out the

possibility that remaining, unobserved characteristics of the respondents may cause

endogeneity. Fixed-effects estimation may serve to overcome this problem; this could

not, however, be implemented in our case as we were relying on cross-sectional data.

Second, survey data may be affected by measurement errors. We reduce this risk by

using thoroughly tested questionnaire formulations wherever possible, using clear and

easily understandable formulations, and ordering the questions in such a way as to limit

unintended influences of one item on another. Third, although we believe that it is more

likely that experiences and expectations affect LS, rather than the other way around, it is

possible that causality may be reversed. It cannot be guaranteed that endogeneity does

not exist in Equation (3) and that this does not cause a bias in the results, especially with

regard to causality. As a consequence, we will be cautious when making statements about

cause and effect relationships.

For regressions, including the expectation variables, we exclude respondents who

do not believe that climate change is occurring, or will occur, in Germany

(approximately 5% of the sample). We include these households in two robustness

checks, however, by assuming neutral expectations (g3 D 0 ), and by adding a sixth category of the expectation variable for respondents who do not expect climate change

to occur in Germany.

Turning to the proposed indirect effects of damage experiences, it should be noted that

the concept of direct and indirect effects is based on the assumption that the relationship

of climate experience and expectations, on the one hand, and LS, on the other, is causal.

As already stated, the limited availability of data means that we are unable to provide

Journal of Environmental Planning and Management 2205

empirical proof of this. We are, however, confident that there are good theoretical

arguments proving that measured effects are indeed causal.

As the first step, climate expectations are estimated (the empirical analogue of

Equation (1b)). The variable of interest is a latent continuous variable capturing climate

expectations (C�i ), which has to be transferred to the observed ordinal variable Ci:

C � i D k C λ�Ei C m�Wi C xi (4a)

Ci D j if vj ¡ 1 < C�i ¡ k�vj; j D 1; :::; 5; v0 D ¡ 1 ; v5 D C 1

(4b)

Wi is a set of control variables which includes Xi from Equation (3) and further variables

which are assumed to correlate with C� but not with LS. These variables include environmental and political attitudes, and information sources for daily news. xi is an

error term. The unobserved variable C�i is transferred to the observed ordinal variable Ci by Equation (4b). vj are the thresholds of the latent variable. As none of the thresholds is

fixed to a value, they incorporate the constant k which must be subtracted from C�i . The functional form of Equation (4a) and the selection of control variables Wi are

derived from literature on climate risk perception or risk awareness, such as Akerlof et al.

(2013), Benight et al. (2007), Kellens et al. (2011), Whitmarsh (2008), van der Linden

(2015), Spence et al. (2011), and Bichard and Kazmierczak (2012). In these studies,

experience is measured using binary variables (i.e. having or not having experienced any

climate-related damage), as we have done in our empirical application. 12

Control

variables which are expected to correlate with climate change expectations and which are

measured in the majority of published studies, including our own, include socio-

demographic variables, political, environmental, and risk attitudes, as well as information

source.

In the following, Equation (3) is estimated by OLS (by ordered probit as a robustness

check), and Equation (4a/4b) is estimated using an ordered probit model.

As the error terms ei and xi may correlate with each other due to unobserved personality traits of the respondents, we estimate the system of Equations (3) and (4a/4b)

separately (by OLS and ordered probit) and simultaneously by the user-written Stata-

command cmp (Roodman 2011). This command enables simultaneous estimation of

coefficients using different estimation techniques and data levels, while taking account of

a possible correlation between the error terms. The simultaneous regression also provides

an estimate for the correlation of the error terms which can be used as an indicator of the

necessity of taking a simultaneous approach.

To check whether an indirect effect of damage experience on SWB can be detected in

our data, the following term must be evaluated (which is the empirical analogue to

Equation (2)):

dLSi

dEi D b C

X5 j D 1

gj � dPrðCi D jÞ

dEi

� � (5)

The indirect effect of Ei on LSi is, therefore, the sum of marginal effects of Ei on the

estimated probabilities that observation i takes the expectation level of j, multiplied by

the marginal effects of these expectation levels on LSi.

2206 D. Osberghaus and J. K€uhling

In the following, we define:

X5 j D 1

gj � dPrðCi D jÞ

dEi

� � D λ

The extended empirical model (Equations (3) and (4a/4b)) serves as a basis for

hypothesis (c):

Hypothesis (c): The effect described in hypothesis (a) ( dLSi dEi

) can be divided into a direct

effect b from the mere damage experience and an indirect effect λ , the LS effect of changing expectations of negative climate impacts, implied by experience (λ < 0 ).

In order to estimate the magnitude and significance of the indirect effect λ , we combine the coefficients from regressions of Equation (3), which is estimated by OLS,

and Equation (4a/4b) which is an ordered probit model. To derive the term dPrðCi D jÞ

dEi , one

might set the covariates at representative values and calculate the marginal effect for this

representative household. For a number of the control variables, however, the choice of

representative value is not obvious. We, therefore, use average marginal effects, which

include all covariates as they are observed. We start with the results taken from separate

regressions and compare these results to the effects implied by the simultaneous

estimation approach.

A further way to assess the magnitude of the indirect effect is the direct comparison of

b yielded by the full model, as formulated in Equation (3), with a restricted model

excluding climate expectations ( P5

j D 1gj�Dij). The coefficient for Ei in the restricted model incorporates the indirect effect of damage experience and should approximately

mirror the total effect (b C λ ).

5. Results

In order to show the pure effects of the control variables, we initially run a regression of

LS without the key variables. We check the robustness of the OLS results by running

ordered probit regressions (not reported in detail. All results not reported in detail are

available on request). As expected, the differences between OLS and ordered probit

estimates are minor (Ferrer-i-Carbonell and Frijters 2004). For the presentation and

discussion of the results, we prefer to use OLS estimates, as these lend themselves to

more intuitive interpretation and also highlight potential differences in the sign and

significance levels of estimated coefficients for key variables. The results of the controls-

only estimation are summarised in Tables 1.3 and 1.4. 13

The significant coefficient

estimates have the expected signs, in particular, the data shows a U-shape effect

according to age, and indicates a positive effect of income and higher education, and

negative effects of unemployment and poor health status. In the ordered probit

regressions, these effects are confirmed.

In the next step, we include damage experience in the estimation. The respective

results regarding the key variables are presented in the column “Model 1” in Table 3. 14

The results show that LS decreases significantly as a result of experience of damage

caused by heat waves. Damage experience resulting from other EWEs shows negative,

although insignificant, effects on LS. This will be further elaborated in Section 6.

“Model 2” in Table 3 focuses on the relationship between damage expectations for the

future and current LS. The coefficients are significant and show the expected signs

(g4;5 < g3 D 0, see Equation (3)).15

Journal of Environmental Planning and Management 2207

T a b le 3 .

O L S re g re ss io n re su lt s. D e p e n d e n t v a ri a b le : li fe

sa ti sf a c ti o n (L S ).

V a ri a b le in

th e m o d e l

V a ri a b le in

th e d a ta

M o d e l 1

M o d e l 2

M o d e l 3

D a m a g e e x p e ri e n c e

(p a st )

P a rt ic ip a n t h a s a lr e a d y e x p e ri e n c e d fi n a n c ia l o r h e a lt h d a m a g e b y th e fo ll o w in g :

H e a t w a v e

¡0 .6 6 2 ��

� (0 .1 5 9 )

� ¡0

.5 5 9 ��

� (0 .1 6 7 )

S to rm

¡0 .0 4 7 5 (0 .0 6 5 5 )

� ¡0

.0 5 0 9 (0 .0 6 8 2 )

H e a v y ra in

¡0 .0 2 2 8 (0 .0 6 2 8 )

� ¡0

.0 2 1 1 (0 .0 6 5 4 )

F lo o d

¡0 .1 6 3 (0 .0 8 5 5 )

� ¡0

.1 2 4 (0 .0 8 9 9 )

C li m a te c h a n g e

e x p e c ta ti o n (f u tu re )

E x p e c te d c o n se q u e n c e s o f c li m a te c h a n g e o n in d iv id u a l li v in g c o n d it io n s in

th e n e x t d e c a d e s

V e ry

p o si ti v e (j D 1 )

� 0 .4 8 1 (0 .8 8 8 )

0 .4 4 6 (0 .8 9 1 )

R a th e r p o si ti v e (j D 2 )

� ¡0

.0 2 5 3 (0 .1 7 7 )

¡0 .0 5 2 3 (0 .1 7 6 )

N e it h e r p o si ti v e n o r

n e g a ti v e (j D 3 )

R e fe re n c e g ro u p

R a th e r n e g a ti v e (j D 4 )

� ¡0

.1 9 2 ��

� (0 .0 5 7 9 )

¡0 .2 0 4 ��

� (0 .0 5 7 9 )

V e ry

n e g a ti v e (j D 5 )

� ¡0

.5 0 9 ��

� (0 .1 4 4 )

¡0 .5 0 3 ��

� (0 .1 4 4 )

C o n tr o l v a ri a b le s

in c lu d e d

O b se rv a ti o n s

4 ,7 6 6

4 ,2 2 3

4 ,1 7 2

R 2

0 .1 4 8

0 .1 5 4

N o te : R o b u st st a n d a rd

e rr o rs in

p a re n th e se s. T h e a st e ri sk s (� , �� , a n d �� ) d e n o te si g n ifi c a n c e le v e ls o f 1 0 % , 5 % , a n d 1 % , re sp e c ti v e ly .

2208 D. Osberghaus and J. K€uhling

“Model 3” includes all variables for damage experience and damage expectations. The

estimated effect of heat wave experience and expectations of general consequences for

personal living conditions continue to diverge significantly different from zero.

Control variables are included in all specifications presented in Table 3. For full

estimation results, see Tables 1.3 and 1.4. Table 1.4 (here the sample remains identical

across specifications) shows that signs, magnitudes, and significance levels of control

variables do not change remarkably where climate variables are included.

In order to identify and quantify an indirect effect of damage experience on LS, we

concentrate on the effect of heat waves (we do not, however, exclude other damage

variables from our estimations).

First, we will evaluate the indirect effect by considering regressions of LS and climate

expectations Ci separately. The LS regression is almost identical to “Model 3” in Table 3.

Minor differences stem from a slight reduction in the sample size, as this results in

missed observations in the estimation of climate damage expectations. Complete results

are reported as “Model 4” in Table 1.3.

Key results from the ordered probit estimation of climate damage expectations

(Equation (4a/4b)) are presented in Table 4. As well as all of the control variables from

the LS regression, further control variables are also included (descriptive statistics see

Table 1.1). The results indicate a correlation of high personal damage expectations with

low household income, non-homeownership, risk aversion, being overweight, pro-

environmental attitudes, left-wing partisanship, not using the internet as a daily source of

information, and damage experience from heat waves (all relations are significant by at

least 10%). Note that the same sample has been used as in the separate LS regression.

The results of the ordered probit model facilitate the calculation of average marginal

effects for each climate damage expectation level (see Table 5, column 2). The marginal

effects show the expected signs, with decreasing probabilities for low expectation levels

and increasing probabilities for higher expectation levels if heat wave damage occurs.

Multiplying these probability changes with the LS effects of the respective expectation

levels (column 3 of Table 5, taken from “Model 4” in Table 1.3), yields the indirect LS

effects of damage experience for each level (column 4 of Table 5), which sum to the total

indirect effect λ (see Equation (5)). The indirect effect of past damage experience on LS implied by the change of future

damage expectations is small, but nonetheless shows significant divergence from zero (p

< 0.01). Compared to the total effect, the indirect effect amounts to approximately 5.1%

of the total effect.

As already mentioned, the indirect effect can also be determined by directly

comparing the coefficients for damage experience yielded by “Model 1” (restricted

model, excluding damage expectations) and “Model 3”. Given that it also contains data

from the same sample, we are also able to draw on the results presented in Table 1.4. The

coefficients for damage experience are ¡0.591 in “Model 1” and ¡0.559 in “Model 3”. The difference is indeed of a similar magnitude as that of the indirect effect calculated

above (¡0.032), and the percentage is 5.4% of the total effect. In the next step, we repeat the two regressions (on LS and damage expectations) in a

simultaneous equation model using the Stata command cmp by Roodman (2011). The

results (Table 1.6) do not confirm a correlation of the error terms, indicating that a

simultaneous estimation of the two regressions is not necessary. However, if it is calculated,

the simultaneous estimation results are similar to those presented above. The indirect effect

(calculated in Table 1.7) is smaller (¡0.023 instead of ¡0.031), but still significantly different from zero (albeit on a lower significance level, p < 0.1). In terms of percentage of

the total effect, the simultaneous estimations imply an indirect effect of around 3.9%.

Journal of Environmental Planning and Management 2209

6. Discussion of results

The results offer new insights into the interrelationships between LS, damage experience

due to EWE, and concern about future climate change (damage expectations). There are,

however, some caveats to be made. Our analysis is primarily based on self-reported

survey data, which implies a risk of inaccurate, strategic, or context-dependent response

behaviour. Furthermore, the dataset does not allow for controlling of measurement error,

for example, by the use of multiple item constructs. The risk of endogeneity has already

been mentioned. These limitations must be kept in mind when interpreting the data. It

should, in particular, be noted that the concept of direct and indirect LS effects of climate

experience is only valid if the measured relationships are causal. Given the data

available, we are not yet able to confirm this. We will discuss the following topics

Table 4. Ordered probit regression results. Dependent variable: climate damage expectations (Ci).

Variable in the model Variable in the data Coefficients

(robust standard errors)

Income Ln of Household income in € ¡0.170�� (0.0442) Homeownership Ownership of the residence ¡0.0837� (0.0441) Health Underweight ¡0.0344 (0.233)

Normal weight Reference group

Overweight 0.0871 �� (0.0422)

Obesity 0.0837 (0.0532)

Personal attitudes Own health status is very important 0.0747 � (0.0436)

Protection of nature and the environment is very important

0.108 ��

(0.0417)

Combatting climate change is very important 0.511 ��

(0.0419)

Stated general time preference (high values: high patience)

¡0.0139� (0.00805)

Stated general willingness to take risks ¡0.0297�� (0.00975) Partisanship of a left wing party 0.104

�� (0.0388)

Agreement with anthropogenic climate change 0.397 ��

(0.0393)

Agreement with building of new coal power plants

¡0.132�� (0.0515)

Information source for daily news: Internet ¡0.0800�� (0.0376) Damage experience

(past) Participant has already experienced financial or

health damage by the following:

Heat wave 0.291 ��

(0.101)

Storm 0.0564 (0.0465)

Heavy rain 0.0224 (0.0451)

Flood ¡0.0413 (0.0622) Further control variables (see Table 1.1) Included

Threshold 1 (v1) ¡4.604 (0.457) Threshold 2 (v2) ¡3.439 (0.434) Threshold 3 (v3) ¡1.350 (0.430) Threshold 4 (v4) 0.390 (0.429)

Observations 3,954

Pseudo-R 2

0.0783

Note: The asterisks ( � , �� , and

�� ) denote significance levels of 10%, 5%, and 1%, respectively.

2210 D. Osberghaus and J. K€uhling

separately: relationship of LS and damage experience, LS and damage expectations, and,

finally, the separation of direct and indirect LS effects of damage experience. Where

possible, we will discuss our results in the light of findings from previous empirical

studies.

6.1. Damage experience and life satisfaction

It has been shown that the reported past experience of having suffered financial losses or

health damage as a result of heat waves correlates significantly with current LS, keeping

everything else equal (“Model 1” and “Model 3” in Table 3). The effect, which is robust

over all specifications, is similar in its magnitude to that of being unemployed. 16

This

result is even more striking given that damage experiences from other EWEs (flooding,

storms, and heavy rain) do not show significant effects on LS. 17

Our results, therefore,

complement specific findings in the heterogeneous literature between EWE experience

and SWB (see Table 1).

Although the data do not allow certain attribution of types of damage (health-related

vs. financial) to kinds of EWEs, we must here presume that heat waves are qualitatively

different from the other events, in that they tend to primarily cause health-related

damage, whereas other events cause mainly financial damage. This assumption is not

insignificant, as there are reports on the effects of flooding on mental health (Fernandez

et al. 2015, WHO 2013). Nevertheless, for a representative household sample in

Germany, we consider this assumption to be fair and so conclude that health-related

climate damages exhibit higher LS effects than financial damage. We recall that health-

related damage was defined by the necessity to consult a doctor, while financial damage

was not restricted by a lower limit. Hence, health-related damage could, in theory, be

more severe than (possibly low) financial damage. Furthermore, material damage may be

more easily compensated either by savings or by insurance companies. 18

In Germany,

there is a private insurance market for storm and hail damage which covers almost all

households. In the case of flooding, the insurance density is lower (around 30%), which

has, on a number of occasions, prompted the government to grant substantial relief

payments. The fact that direct financial compensation can generally be given for material

damage, but not for health-related effects, constitutes an important difference between

Table 5. Calculation of the indirect effect of damage experience on LS.

Climate damage expectation levels (Ci)

Average marginal effect (change of estimated probability

with regard to heat wave experience) LS effect of

expectation level Indirect effects

Ci D 1 ¡0.00121� (0.000638) 0.427 (0.901) ¡0.000516 (0.00108) Ci D 2 ¡0.0154�� (0.00555) ¡0.0277 (0.182) 0.000428 (0.00281) Ci D 3 ¡0.0880�� (0.0304) 0 (reference group) 0 Ci D 4 0.0727�� (0.0251) ¡0.218�� (0.0596) ¡0.0158�� (0.00433) Ci D 5 0.0320�� (0.0112) ¡0.469�� (0.150) -0.0150�� (0.00294) Sum over

expectation levels (λ )

n.a. n.a. -0.0309 ��

(0.00713)

Note: The asterisks ( � , �� , and

�� ) denote significance levels of 10%, 5%, and 1%, respectively. Robust standard

errors of marginal effects and coefficients in parentheses. Standard errors in column 4 have been calculated manually using the error propagation formulas given in Taylor (1997).

Journal of Environmental Planning and Management 2211

the two types of damage. Another explanation focuses on the temporal dimension. It is

possible that health-related damage has longer term effects on LS. Extreme heat exposure

may aggravate several chronic diseases, including cardiovascular, respiratory, renal, and

gastroenterology diseases (Centres for Disease Control and Prevention 2010; Hansen

et al. 2008; Manser et al. 2013). Hospital admissions indeed increase during heat waves

in Germany, as shown by Zierbarth, Schmitt, and Karlsson (2014). In comparison,

financial damage may have only temporary implications for LS. Given that we do not

know when the damage in our sample occurred, however, we are unable to provide

certain confirmation of this. In our view, it can be assumed that the missing (significant)

effect of financial damage is partly due to this discounting effect, whereas health-related

damage may have longer lasting LS effects. We do, however, see scope for further

research into the temporary dimensions of LS effects of EWEs. Another caveat to be

made in regard to the dataset is the nature of damage experience data (self-reported

instead of objective data), and the limited information available concerning the health

status of participants. Health status can, of course, be captured to some extent by the

control variables BMI, outdoor activities, and risk aversion.

6.2. Damage expectations and life satisfaction

The relationship between expectations regarding future climate change impacts and current

LS was analysed using ordinal data from a five-point Likert scale questionnaire item. It

should be noted that variables capturing expectations by surveys may be highly context

dependent and influenced by factors for which we can only partially control. We identify

the following relationship in our data; those participants who expect climate change to

have adverse effects on individual living conditions in the coming decades tend to be less

satisfied (“Model 2” in Table 3). The strength of the correlation is a bit lower than the

effect of damage experience, but highly significant. This means that concerns about the

impact of future climate change on personal living conditions affects LS, even today.

6.3. Direct and indirect LS-Effects of damage experience

As described in Section 1, previous literature has demonstrated that experiences of

climate-induced EWEs may influence LS (see Table 1). The explicit inclusion of damage

expectations for the future facilitates deeper analysis of this relationship. Our theoretical

model and empirical results suggest that the LS effect of damage experiences can indeed

be divided into direct effects, a result of the mere loss experienced in the past, and a

significant indirect effect in terms of changed damage expectations for the future. This

means that hypothesis (c), stated in the theory part of this paper, is not disproved by our

data. Having said this, estimations of the indirect effect indicate that, although direct and

indirect effects are significant, the indirect effect is very small compared to the total

effect (around 5%). This suggests that the LS effect of climate damage experience stems

mainly from the past damage experience itself and only to a small extent, from the

experiences which drive a change in future damage expectations.

6.4. Estimation of climate damage expectations

In order to estimate the effect of heat wave experience on climate expectations, an

ordered probit regression has been conducted which, beside extreme weather experience,

2212 D. Osberghaus and J. K€uhling

also includes all controls of the LS regressions and a number of additional variables. The

results shall here be briefly reviewed.

Our data suggest that individuals with high personal climate damage expectations

tend to exhibit the following characteristics: politically, they are partisans of left-wing

parties. Such participants tended to have strong pro-environmental attitudes, rating

environmental protection, and the combat against global climate change as very

important, stating that climate change is mainly induced by man, and expressing a dislike

of new coal power plants. Furthermore, they are generally risk averse, have a higher than

normal BMI, and prefer to use news sources other than the internet. Economically, they

are poorly situated, with relatively low income and no homeownership.

Apart from heat waves, damage experiences with EWEs do not seem to have a

significant effect on damage expectations. This contrasts with the results from studies

carried out by Bichard and Kazmierczak (2012) amongst flood-prone households in the

UK, which report higher climate change risk perceptions of previously flooded

households, and that of Spence et al. (2011), in which respondents in the UK, who have

experienced flooding in their local area (not necessarily in their own homes), report

higher concerns regarding climate change. As already mentioned, Whitmarsh (2008)

reports insignificant correlations between flooding experience and personal climate risk

perception. Two studies where climate change experience is measured (explicitly or

implicitly) as heat experience find positive and significant correlations with risk

perception (Akerlof et al. [2013] for Alger County, MI, USA, and van der Linden [2015]

for the UK). A direct comparison of the results is hampered, however, by the fact that

most studies use slightly different formulations and measures of risk perception and

experience.

Our data do not confirm an effect of further socio-demographic variables on climate

expectations, such as sex, education, occupation, or family status.

In regard to political and environmental attitudes, our data broadly confirm the

empirical results from previous studies on the determinants of climate risk perception

(Brody et al. 2008; Leiserowitz 2006; Liu, Smith, and Safi 2014; Owen et al. 2012; Safi,

Smith, and Liu 2012; van der Linden 2015; Whitmarsh 2008). 19

The significant negative

effect of income on the extent of concerns is not, however, noted in the majority of

previous studies.

7. Conclusions

This paper deals with the triangular interrelationships of damage experience in the past,

climate-change-induced damage expectations for the future, and current SWB. In

particular, the following research questions are addressed. How does damage experience

in the past affect current SWB? How do damage expectations for the future relate to

current SWB? Is there an indirect effect of damage experience as changed damage

expectations affect SWB and, what is the extent of any possible effect? The latter two

questions, in particular, have been rarely addressed in literature.

To answer these questions, we utilise data from a new large-scale survey carried out

amongst German households and perform various regression analyses.

The results can be summarised as follows: we find a strong and significant relationship

between heat wave damage experience and current SWB, whereas damage experience

due to other EWEs (storms, hail and heavy rain, and floods) does not have significant

effects. This finding may be explained in several different ways, with explanations

ranging from the possibility of insuring against material damage, but not against health

Journal of Environmental Planning and Management 2213

damage, to a discounting effect which is relevant to material damage rather than health-

related effects. We also find a significant and robust relationship between climate-

change-induced damage expectations in the future with current SWB. To our knowledge,

this is the first analysis which draws links between climate-change-induced damage

expectations and current SWB. Furthermore, the SWB effect of experiences can be

separated into a direct effect from the mere damage event and a small but significant

indirect effect which affects current SWB as a result of changed damage expectations for

the future. The estimated ratio of this indirect effect over the total effect is around 5%. As

an important limitation, it should be noted that endogeneity cannot be ruled out in some

of the estimations. Causality, although theoretically quite plausible, cannot, therefore, be

empirically backed using the available data. The results suggest several directions for

further research. First, the strong and robust effect of heat waves on SWB (as such, but

also compared to the non-significant effects of other weather events) deserves closer

analysis. One possibility is to analyse whether insurance coverage is able to mitigate or

even offset SWB effects of financial weather damage and/or how quickly individuals

adjust to material damage (discounting effect). Second, the relationship between

individual SWB and climate change expectations for the future seems to be a relevant

factor, not least for the acceptance of climate policies, which has yet to be adequately

studied. Ultimately, differences between countries might be established by means of

international analyses of this relationship. We believe the present study to be an initial

step in this regard.

Acknowledgements

We would like to thank J€urgen Bitzer, Klaus Eisenack, Katherine von Graevenitz, David Maddison, as well as Heinz Welsch for useful comments and fruitful discussions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was supported by the German Ministry for Education and Research (BMBF) under Grant 01LA1113C. The funding source had no involvement in study design, collection, analysis and interpretation of the data, writing of the report, and the decision to submit the article for publication.

Notes

1. Welsch and K€uhling (2009) give an overview about some of the following EWE papers and discuss them in more detail in an environmental valuation framework using the happiness approach. An update of the literature review is available in Welsch and Ferreira (2014).

2. We refrain from calculating monetary values since, in survey data such as ours, the estimated income coefficient (marginal utility of income) tends to be underestimated. Hence, an estimate of the monetary value of the EWE effect would be biased upwards. Nevertheless, the monetarisation of EWE experience and expectations is a promising field of further research.

3. Participating households tend to be larger than households in a representative German sample (ratio of single households 25.8% vs. 40.4%). In terms of other comparable household statistics (e.g. income, federal states), our sample is fairly representative for Germany. As only heads of households participated in the survey, the statistics for individual characteristics differ from the general population in Germany. The sample, in relation to the general

2214 D. Osberghaus and J. K€uhling

population, is characterised by more males, higher age, higher education, and less risk aversion (Osberghaus 2015).

4. As expected, the correlation matrix shows that SWB (life satisfaction) is negatively correlated with both damage experiences and adverse climate expectations. Variables of damage experiences and variables of adverse climate expectations, respectively, are significantly positively correlated with each other indicating, for example, that a storm can cause heavy rain which, in turn, can result in flooding. It is, therefore, possible that the respondents refer to the same event under the three categories. Such a relationship does not exist for the variable, heat waves, which shows especially significant correlations and stands in the focus of the analysis.

5. This and other questions, which are relevant for eliciting the key variables, are available in Table 1.5.

6. From here on, we will use “Life Satisfaction” (LS) and SWB interchangeably. 7. In a comparable cross section of more than 19,500 individuals in Germany (German socio-

economic panel of 2012), there exists a high correlation between an aggregated LS variable basing on 11 domains of life and the single-item LS variable (Spearman’s rho D 0.64, p < 0.01).

8. As we do not know where the participants experienced the damage, we cannot correlate the subjective experience with objective weather data. Evidence for different surveys in the USA suggests that weather events reported by individuals correlate fairly well with actual weather data (Akerlof et al. 2013; Howe et al. 2014, Ruddell et al. 2012). Nevertheless, this limitation has to be taken into account for the interpretation of results.

9. The question on climate change expectations was posed after the questions on climate change experience. Hence, we cannot exclude an unintended influence of responding to the latter on the responses to the former. However, the participants spent several minutes in between answering questions about unrelated topics; we therefore expect any influence to be minor.

10. “Don’t know” answers are omitted in the analyses. 11. As a robustness check, we include a dummy variable indicating those respondents who

evaluate each of six different global challenges (ranging from environmental to social and economic topics) as “very important”, in order to capture a rough measure of pessimism. Descriptive data are available upon request.

12. None of the cited studies discuss the implications of using a binary variable for experience, instead of measures for severity or time of the experienced event. We assume that, as in our case, this procedure is due to limited data availability.

13. Table 1.3 depicts estimation results with all available observations per specification. Table 1.4 shows the results with a reduced sample as it is available in the most comprehensive specification (Model 3). There are no large differences with regard to signs and significance levels of the estimates.

14. Complete results of the regressions are presented in Table 1.3 (full sample) and Table 1.4. (reduced sample).

15. Including the dummy variable capturing general pessimism or including respondents who do not believe in climate change, as described in Section 4, does not change the results.

16. The inclusion of various interaction variables (age, sex, health status, farmer households) showed no significant interaction with the LS effect of experiencing heat wave damage. Figure 1.1 shows that Germany’s summer mean temperatures have indeed varied considerably in the last 20 years. Locally, the variations were even larger.

17. There is still no significant LS effect of damage experience when flooding, storm, and heavy rain damage experiences are aggregated to one variable.

18. In contrast to our results Luechinger and Raschky 2009�find a negative impact of floods on life satisfaction that is sizeable, robust and significant.� They also; �find that risk transfer mechanisms, such as mandatory insurance, have large mitigating effects.� Capturing their latter finding and our reasoning of compensation through insurance companies, the discrepancies between both studies may arise from the samples used. While Luechinger and Raschky 2009 use data from 16 European countries between 1973 and 1998, our dataset corresponds to German households in the year 2012. Apart from general differences in the interviewed household samples, one may assume better functioning insurance markets in Germany in recent years, which may explain the seemingly contradictory results. For a more

Journal of Environmental Planning and Management 2215

detailed insight into the ‘value individuals place on the reduction of increased flood risks by insurance coverage’ see Botzen and van den Bergh 2009, 2012a, 2012b.

19. Stemming from different scientific communities, the mentioned studies include various statistical methods, variable formulations, and additional determinants. For instance, van der Linden (2015) examines the relationships of sociocultural factors (which are not in our dataset) with climate change risk perception and reports relatively high shares of variance explained by these factors.

References

Akerlof, K., E.W. Maibach, D. Fitzgerald, A.Y. Cedeno, and A. Neuman. 2013. “Do People “Personally Experience” Global Warming, and If So How, and Does It Matter?” Global Environmental Change 23 (1): 81�91. doi:http://dx.doi.org/1.1016/j.gloenvcha.2012.07.006

Ambrey, C.L., and C.M. Fleming. 2011. The Influence of Natural Environment and Climate on Life Satisfaction in Australia. Discussion Paper No. 2011-01, Brisbane: Griffith University. Discussion Paper No. 2011�01, Griffith University, Brisbane.

Becchetti, L., S. Castriota, and L.B. D. Andr�es. 2007. Climate Happiness and the Kyoto Protocol: Someone Does Not Like It Hot. Departmental Working Paper, Rome: University of Rome Tor Vergata. Departmental Working Paper, University of Rome Tor Vergata, Rome.

Benight, C., E. Gruntfest, M. Hayden, and L. Barnes. 2007. “Trauma and Short-Fuse Weather Warning Perceptions.” Environmental Hazards 7 (3): 220�226. doi:10.1016/j.envhaz.2007. 07.002

Bichard, E., and A. Kazmierczak. 2012. “Are Homeowners Willing to Adapt to and Mitigate the Effects of Climate Change?” Climatic Change 112 (3�4): 633�654. doi:1.1007/s10584-011- 0257-8

Botzen, W., and J.C.J.M. van den Bergh. 2009. “Bounded Rationality, Climate Risks, and Insurance: Is There a Market for Natural Disasters?” Land Economics 85 (2): 265�278.

Botzen, W., and J.C.J.M. van den Bergh (2012a). “Risk Attitudes to Low-Probability Climate Change Risks: WTP for Flood Insurance.” Journal of Economic Behavior and Organization 82: 151�166.

Botzen, W., and J.C.J.M. van den Bergh (2012b). “Monetary Valuation of Insurance Against Flood Risk Under Climate Change.” International Economic Review 53 (3): 1005�1025.

Brereton, F., J.P. Clinch, and S. Ferreira. 2008. “Happiness, Geography and the Environment.” Ecological Economics 65 (2): 386�396.

Brody, S.D., S. Zahran, A. Vedlitz, and H. Grover. 2008. “Vulnerability and Public Perceptions of Global United States.” Environment and Behavior 40 (1): 72�95.

Carroll, N., P. Frijters, and M. Shields. 2009. “Quantifying the Costs of Drought: New Evidence from Life Satisfaction Data.” Journal of Population Economics 22 (2): 445�461.

Centres for Disease Control and Prevention. 2010. Climate and Health. Atlanta, GA: CDC. http:// www.cdc.gov/climateandhealth/effects/

Cu~nado, J., and F.P. de Gracia. 2013. “Environment and Happiness: New Evidence from Spain.” Social Indicators Research 112: 549�567.

Diener, E., R.A. Emmons, R.J. Larsen, and S. Griffin. 1985. “The Satisfaction with Life Scale.” Journal of Personality Assessment 49 (1): 71�75.

DWD (Deutscher Wetterdienst). 2014. “Zeitreihen von Gebietsmitteln.” Accessed July 15, 2014. http://www.dwd.de/bvbw/appmanager/bvbw/dwdwwwDesktop?_nfpbDtrue&_pageLabelD_ dwdwww_klima_umwelt_klimadaten_deutschland&T82002gsbDocumentPathDNavigation%2 FOeffentlichkeit%2FKlima__Umwelt%2FKlimadaten%2Fkldaten__kostenfrei%2Fdaten__ge bietsmittel__node.html%3F__nnn%3Dtrue.

Feddersen, J.R., R. Metcalfe, and M. Wooden. 2012. Subjective Well-being: Weather Matters, Climate Doesn’t. Melbourne Institute Working Paper No. 25 Melbourne: University of Melbourne.

Fernandez, A., J. Black, M. Jones, L. Wilson, L. Salvador-Carulla, T. Astell-Burt, and D. Black. 2015. “Flooding and Mental Health: A Systematic Mapping Review.” Plos One 10 (4): 1�20. doi:10.1371/journal.pone.0119929

Ferreira, S., and M. Moro. 2010. “On the Use of Subjective Well-being Data for Environmental Valuation.” Environmental and Resource Economics 46 (3): 249�273.

Ferrer-i-Carbonell, A., and P. Frijters. 2004. “How Important Is Methodology for the Estimates of the Determinants of Happiness?” The Economic Journal 114: 641�659.

2216 D. Osberghaus and J. K€uhling

http://dx.doi.org/1.1016/j.gloenvcha.2012.07.006

http://dx.doi.org/10.1016/j.envhaz.2007.07.002

http://www.cdc.gov/climateandhealth/effects/

http://www.dwd.de/bvbw/appmanager/bvbw/dwdwwwDesktop?_nfpb=true&_pageLabel=_dwdwww_klima_umwelt_klimadaten_deutschland&T82002gsbDocumentPath=Navigation%2FOeffentlichkeit%2FKlima__Umwelt%2FKlimadaten%2Fkldaten__kostenfrei%2Fdaten__gebietsmittel__node.html%3F__nnn%3Dtrue

http://dx.doi.org/10.1371/journal.pone.0119929

Ferrer-i-Carbonell, A., and J.M. Gowdy. 2007. “Environmental Degradation and Happiness.” Ecological Economics 60: 509�516.

Field, C.B., V. Barros, T.F. Stocker, Q. Dahe, D.J. Dokken, K.L. Ebi, and M.D. Mastrandrea. 2012. Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. Cambridge: Cambridge University Press.

Frey, B.S., and A. Stutzer. 2002. “What Can Economists Learn from Happiness Research?” Journal of Economic Literature 40: 402�435.

Frijters, P., and B.M.S. van Praag. 1998. “The Effects of Climate on Welfare and Well-being in Russia.” Climatic Change 39 (1): 61�81.

Gr€un, C., and N. Grunewald. 2010. “Subjective Well-being and the Impact of Climate Change.” Paper presented at the German Development Conference, Hannover, Germany, June 18�19.

Hansen, A.L., P. Bi, P. Ryan, M. Nitschke, D. Pisaniello, and G. Tucker. 2008. “The Effect of Heat Waves on Hospital Admissions for Renal Disease in a Temperate City of Australia.” International Journal of Epidemiology 37 (6): 1359�1365. doi:10.1093/ije/dyn165

Howe, P.D., H. Boudet, A. Leiserowitz, and E.W. Maibach. 2014. “Mapping the Shadow of Experience of Extreme Weather Events.” Climatic Change 127: 381�389. doi:10.1007/ s10584-014-1253-6

IPCC. 2014. Climate Change 2014: The Physical Science Basis Summary for Policymakers. Geneva: Intergovernmental Panel on Climate Change.

Kellens, W., R. Zaalberg, T. Neutens, W. Vanneuville, and P. De Maeyer. 2011. “An Analysis of the Public Perception of Flood Risk on the Belgian Coast.” Risk Analysis 31 (7): 1055�1068. doi:10.1111/j.1539-6924.2010.01571.x

Kimball, M., H. Levy, F. Ohtake, and Y. Tsutsui. 2006. “Unhappiness after Hurricane Katrina.” Working Paper 12062. Cambridge: National Bureau of Economic Research.

Kountouris, Y., and K. Remoundou. 2011. “Valuing the Welfare Cost of Forest Fires: A Life Satisfaction Approach.” Kyklos 64 (4): 556�578.

Leiserowitz, A. 2006. “Climate Change Risk Perception and Policy Preferences: The Role of Affect, Imagery, and Values.” Climatic Change 77 (1�2): 45�72. doi:10.1007/s10584-006- 9059-9

Liu, Z., W.J. Smith, and A.S. Safi. 2014. “Rancher and Farmer Perceptions of Climate Change in Nevada, USA.” Climatic Change 122 (1�2): 313�327. doi:10.1007/s10584-013-0979-x

Luechinger, S., and P.A. Raschky. 2009. “Valuing Flood Disasters Using the Life Satisfaction Approach.” Journal of Public Economics 93 (3�4): 620�633.

Mackerron, G., and S. Mourato. 2013. “Happiness is Greater in Natural Environments.” Global Environmental Change 23 (5): 992�1000.

Maddison, D., and K. Rehdanz. 2011. “The Impact of Climate on Life Satisfaction.” Ecological Economics 70 (12): 2437�2445.

Manser, C.N., M. Paul, G. Rogler, L. Held, and T. Frei. 2013. “Heat Waves, Incidence of Infectious Gastroenteritis, and Relapse Rates of Inflammatory Bowel Disease: A Retrospective Controlled Observational Study.” The American Journal of Gastroenterology 108 (9): 1480�1485. http:// dx.doi.org/10.1038/ajg.2013.186

Murray, T., D. Maddison, and K. Rehdanz (2013), “Do Geographical Variations in Climate Influence Life Satisfaction?” Climate Change Economics 4 (1): 1350004. doi: 10.1142/S2010007813500048

Osberghaus, D., C. Schwirplies, and A. Ziegler. 2013. Klimawandel in Deutschland: Risikowahrnehmung, Wissensstand und Anpassung in privaten Haushalten. Mannheim: ZEW. http://www.rwi-essen.de/forschung-und-beratung/umwelt-und-ressourcen/projekte/eval-map/ ergebnisse/

Osberghaus, D. 2015. “The Determinants of Private Flood Mitigation Measures in Germany � Evidence from a Nationwide Survey.” Ecological Economics 110: 36�50.

Owen, A.L., E. Conover, J. Videras, and S. Wu. 2012. “Heat Waves, Droughts, and Preferences for Environmental Policy.” Journal of Policy Analysis and Management 31 (3): 556�577.

Rehdanz, K., and D. Maddison. 2005. “Climate and Happiness.” Ecological Economics 52 (1): 111�125.

Roodman, D. 2011. “Fitting Fully Observed Recursive Mixed-Process Models with cmp.” The Stata Journal 11 (2): 159�206.

Ruddell, D., S.L. Harlan, S. Grossman-Clarke, and G. Chowell. 2012. “Scales of Perception: Public Awareness of Regional and Neighborhood Climates.” Climatic Change 111 (3�4): 581�607. doi:10.1007/s10584-011-0165-y

Journal of Environmental Planning and Management 2217

http://dx.doi.org/10.1093/ije/dyn165

http://dx.doi.org/10.1007/s10584-014-1253-6

http://dx.doi.org/10.1111/j.1539-6924.2010.01571.x

http://dx.doi.org/10.1007/s10584-006-9059-9

http://dx.doi.org/10.1007/s10584-013-0979-x

http://dx.doi.org/10.1038/ajg.2013.186

http://dx.doi.org/10.1142/S2010007813500048

http://www.rwi-essen.de/forschung-und-beratung/umwelt-und-ressourcen/projekte/eval-map/ergebnisse/

http://dx.doi.org/10.1007/s10584-011-0165-y

Safi, A.S., W.J. Smith, and Z. Liu. 2012. “Rural Nevada and Climate Change: Vulnerability, Beliefs, and Risk Perception.” Risk Analysis 32 (6): 1041�1059. doi:10.1111/j.1539- 6924.2012.01836.x

Spence, A., W. Poortinga, C. Butler, and N.F. Pidgeon. 2011. “Perceptions of Climate Change and Willingness to Save Energy Related to Flood Experience.” Nature Climate Change 1: 46�49. doi:10.1038/nclimate1059

Taylor, J.R. 1997. An Introduction to Error Analysis (2nd ed.). Sausalito, CA: University Science Books.

van der Linden, S. 2015. “The Social-Psychological Determinants of Climate Change Risk Perceptions: Towards a Comprehensive Model.” Journal of Environmental Psychology 41: 112�124. doi:10.1016/j.jenvp.2014.11.012

Welsch, H., and S. Ferreira. 2014. Environment, Well-being, and Experienced Preference, Discussion Paper No. V-367-14, Oldenburg: University of Oldenburg.

Welsch, H., and J. K€uhling. 2009. “Using Happiness Data for Environmental Valuation: Issues and Applications.” Journal of Economic Surveys 23 (2): 385�406.

Welsch, H., and J. K€uhling. 2010. “Pro-Environmental Behavior and Rational Consumer Choice: Evidence from Surveys of Life Satisfaction.” Journal of Economic Psychology 31: 405�420.

Whitmarsh, L. 2008. “Are Flood Victims More Concerned About Climate Change than Other People? The Role of Direct Experience in Risk Perception and Behavioural Response.” Journal of Risk Research 11 (3): 351�374.

WHO (World Health Organization). 2013. Floods in the WHO European Region: Health Effects and their Prevention. http://www.euro.who.int/__data/assets/pdf_file/0020/189020/e96853.pdf

Ziebarth, N.R., M. Schmitt, and M. Karlsson. 2014. “The Short-Term Population Health Effects of Weather and Pollution: Implications of Climate Change.” SOEP Paper No. 646. Berlin: DIW.

Appendix

Figure 1.1. Mean air temperature in June, July, and August between 1990 and 2012 in Germany. Source: DWD (2014).

2218 D. Osberghaus and J. K€uhling

http://dx.doi.org/10.1111/j.1539-6924.2012.01836.x

http://dx.doi.org/10.1038/nclimate1059

http://dx.doi.org/10.1016/j.jenvp.2014.11.012

http://www.euro.who.int/__data/assets/pdf_file/0020/189020/e96853.pdf

Table 1.1. Descriptive statistics of control variables. Control variables for LS regressions are included in Xi; all control variables are included in Wi.

Variable in the model Variable in the data

Mean / in case of binary variables:

share of 1 Min. Max. Obs. Incl. in Xi

Age Age 50.6 18 87 6,404 Yes

Sex Sex 0.324 0 (male) 1 (female) 6,404 Yes

Family status

Married, living together 0.524 0 (no) 1 (yes) 6,341 Yes

Married, living separately 0.032

Single 0.288

Divorced 0.123

Widowed 0.033

Successors Children 0.660 0 (no) 1 (yes) 5,994 Yes

Grandchildren 0.219 6,046

Education Graduated from “Hauptschule” or not graduated

0.147 0 (no) 1 (yes) 6,016 Yes

Graduated from “Realschule” or rest

0.376

Graduated from high school or university

0.477

Occupation Full-time employed 0.603 0 (no) 1 (yes) 5,967 Yes

Part-time employed 0.139

Retired 0.220

Unemployed, searching for employment

0.014

Housewife/-husband 0.005

Other unemployed 0.019

Income Ln of household income in € 7.824 5.521 8.657 5,186 Yes

Homeownership Ownership of the residence 0.555 0 (no) 1 (yes) 6,182 Yes

Personal attitudes Own financial situation is very important

1 0.478 0 (no) 1 (yes) 6,396 Yes

Own health status is very important

1 0.620 6,398 Yes

Protection of nature and the environment is very important

1 0.426 6,397 Yes

Security from crimes is very important

1 0.490 6,397 Yes

Combatting climate change is very important

1 0.522 6,389 No

Stated general time preference (high values: high patience)

6.883 1 11 6,394 Yes

Stated general willingness to take risks

5.826 1 11 6,394 Yes

Stated willingness to take risks regarding own health

4.394 1 11 6,392 Yes

Partisanship of a left wing party 0.390 0 (no) 1 (yes) 5,990 No

Agreement with statement “Humans are mainly responsible for climate change”

0.419 6,007 No

(continued)

Journal of Environmental Planning and Management 2219

Table 1.1. (Continued )

Variable in the model Variable in the data

Mean / in case of binary variables:

share of 1 Min. Max. Obs. Incl. in Xi

Agreement with building of new coal power plants

0.189 6,251 No

Information source for daily news: Internet

0.523 6,004 No

Federal state 16 dummy variables for each state

n.a. 0 1 6,404 Yes

Health BMI less than 18.5 (underweight) 2

0.006 0 (no) 1 (yes) 5,713 Yes

BMI between 18.5 and 25 (normal)

2 0.374

BMI between 25 and 30 (overweight)

2 0.435

BMI higher than 30 (obesity) 2

0.184

Daily outdoor leisure activities 0.301 0 (no) 1 (yes) 6,256 Yes

1 The original questions have five response categories and range from “not important at all” to “very important”. We rescaled them to binary variables capturing the share of “very important”-responses, as we cannot assume an equidistant scale of these ordinal variables. 2 The Body Mass Index (BMI) is calculated by self-reported height and weight. Missing answers are treated as missing and not included in the analysis. According to data of the German socioeconomic panel, the BMI variables correlate significantly with other self-reported health variables, such as the current status of general health, vitality, and physical functioning.

Table 1.2. Correlation matrix of key variables

No. Variable 1 2 3 4 5 6

1 Self-rated life satisfaction (LS) 1.00 � � � � � 2 Damage experience: heat wave ¡0.07 1.00 � � � � 3 Damage experience: storm ¡0.00 0.14 1.00 � � � 4 Damage experience: heavy rain ¡0.00 0.11 0.35 1.00 � � 5 Damage experience: flood ¡0.02 0.06 0.17 0.26 1.00 � 6 Expected consequences of

climate change on individual living conditions in the next decades

¡0.10 0.05 0.01 ¡0.01 ¡0.02 �

Note: Spearman correlation coefficients. Entries in bold denote coefficients which are significantly different from zero (p < 0.01).

2220 D. Osberghaus and J. K€uhling

T a b le 1 .3 .

O L S re g re ss io n re su lt s. D e p e n d e n t v a ri a b le : li fe

sa ti sf a c ti o n (L S ). S a m p le si z e v a ri e s a c c o rd in g to

a v a il a b le o b se rv a ti o n s.

V a ri a b le

in th e m o d e l

C o e ffi c ie n ts (r o b u st st a n d a rd

e rr o rs )

V a ri a b le in

th e d a ta

C o n tr o ls o n ly

M o d e l 1

M o d e l 2

M o d e l 3

M o d e l 4

A g e

¡0 .0 8 3 5 �� (0 .0 1 6 7 )

¡0 .0 8 2 6 �� (0 .0 1 6 8 )

¡0 .0 8 7 1 (0 .0 1 7 8 )

¡0 .0 8 6 5 �� (0 .0 1 7 9 )

¡0 .0 8 8 6 �� (0 .0 1 8 4 )

A g e 2

0 .0 0 0 8 �� (0 .0 0 0 2 )

0 .0 0 0 9 �� (0 .0 0 0 2 )

0 .0 0 0 8 �� (0 .0 0 0 2 )

S e x

S e x (1

D fe m a le )

0 .2 3 3 �� (0 .0 6 4 8 )

0 .2 4 5 �� (0 .0 6 4 8 )

0 .2 1 5 �� (0 .0 6 9 2 )

0 .2 3 0 �� (0 .0 6 9 1 )

0 .2 5 6 �� (0 .0 7 1 1 )

F a m il y st a tu s

M a rr ie d , li v in g to g e th e r

¡0 .0 0 6 8 (0 .0 9 2 5 )

¡0 .0 1 4 2 (0 .0 9 2 8 )

0 .0 5 6 7 (0 .0 9 8 9 )

0 .0 4 9 2 (0 .0 9 9 0 )

0 .1 1 0 (0 .1 0 3 )

M a rr ie d , li v in g se p a ra te ly

¡0 .0 0 6 9 (0 .1 5 9 )

¡0 .0 0 0 9 (0 .1 5 8 )

0 .0 4 5 4 (0 .1 6 8 )

0 .0 5 0 5 (0 .1 6 8 )

0 .0 6 0 5 (0 .1 7 4 )

S in g le

R e fe re n c e g ro u p

D iv o rc e d

¡0 .1 8 5 � (0 .1 1 2 )

¡0 .1 6 8 (0 .1 1 2 )

¡0 .1 4 7 (0 .1 1 9 )

¡0 .1 2 6 (0 .1 2 0 )

¡0 .1 2 2 (0 .1 2 4 )

W id o w e d

¡0 .1 7 1 (0 .1 7 9 )

¡0 .1 6 0 (0 .1 8 1 )

¡0 .3 0 4 (0 .1 8 9 )

¡0 .2 9 6 (0 .1 9 2 )

¡0 .2 4 3 (0 .1 9 7 )

S u c c e ss o rs

C h il d re n (1

D Y e s)

0 .0 4 9 0 (0 .0 7 7 1 )

0 .0 6 3 0 ) (0 .0 7 7 4 )

0 .0 2 9 4 ) (0 .0 8 1 4 )

0 .0 3 9 1 (0 .0 8 1 7 )

0 .0 1 2 2 (0 .0 8 4 5 )

G ra n d c h il d re n (1

D Y e s)

0 .0 8 7 6 (0 .0 7 8 0 )

0 .1 0 1 (0 .0 7 8 2 )

0 .0 7 2 5 (0 .0 8 2 1 )

0 .0 8 9 5 (0 .0 8 2 3 )

0 .0 7 1 8 (0 .0 8 4 6 )

E d u c a ti o n

G ra d u a te d fr o m

“ H a u p ts c h u le ” o r n o t

g ra d u a te d

0 .0 8 5 0 (0 .0 8 4 3 )

0 .0 6 9 1 (0 .0 8 4 1 )

0 .0 8 0 3 (0 .0 9 0 7 )

0 .0 6 5 2 (0 .0 9 0 4 )

0 .1 2 2 (0 .0 9 2 5 )

G ra d . fr o m

“ R e a ls c h u le ” o r

re st

R e fe re n c e g ro u p

G ra d . fr o m

h ig h sc h o o l o r

u n iv e rs it y

0 .1 4 6 �� (0 .0 6 1 3 )

0 .1 4 2 �� (0 .0 6 1 5 )

0 .1 1 2 � (0 .0 6 4 7 )

0 .1 0 5 (0 .0 6 4 9 )

0 .1 0 5 (0 .0 6 6 8 )

O c c u p a ti o n

F u ll -t im

e e m p lo y e d

R e fe re n c e g ro u p

P a rt -t im

e e m p lo y e d

0 .0 0 9 4 7 (0 .0 8 9 0 )

0 .0 2 4 1 (0 .0 8 8 9 )

0 .0 6 0 6 (0 .0 9 3 8 )

0 .0 6 6 5 (0 .0 9 3 8 )

0 .0 5 2 7 (0 .0 9 7 7 )

R e ti re d

0 .3 2 8 �� (0 .1 0 2 )

0 .3 2 8 �� (0 .1 0 1 )

0 .3 3 4 �� (0 .1 1 0 )

0 .3 2 2 �� (0 .1 0 9 )

0 .3 3 6 �� (0 .1 1 1 )

U n e m p lo y e d , se a rc h in g fo r

e m p lo y m e n t

¡0 .6 1 8 �� (0 .2 6 8 )

¡0 .5 9 4 �� (0 .2 6 5 )

¡0 .4 7 2 � (0 .2 7 1 )

¡0 .4 5 6 � (0 .2 6 8 )

¡0 .4 4 5 (0 .2 7 9 )

H o u se w if e /- h u sb a n d

0 .0 1 3 7 (0 .4 0 1 )

¡0 .0 2 1 4 (0 .4 0 1 )

¡0 .0 1 9 1 (0 .4 4 7 )

¡0 .0 5 8 2 (0 .4 4 7 )

¡0 .0 6 2 0 (0 .4 7 3 )

O th e r u n e m p lo y e d , n o t

se a rc h in g fo r

e m p lo y m e n t

0 .3 1 3 (0 .2 2 7 )

0 .3 4 3 (0 .2 2 9 )

0 .2 2 6 (0 .2 4 1 )

0 .2 4 9 (0 .2 4 5 )

0 .1 9 9 (0 .2 5 2 )

(c o n ti n u e d )

Journal of Environmental Planning and Management 2221

T a b le 1 .3 .

(C o n ti n u e d )

V a ri a b le

in th e m o d e l

C o e ffi c ie n ts (r o b u st st a n d a rd

e rr o rs )

V a ri a b le in

th e d a ta

C o n tr o ls o n ly

M o d e l 1

M o d e l 2

M o d e l 3

M o d e l 4

In c o m e

L n o f H o u se h o ld

in c o m e

in €

0 .7 7 7 �� (0 .0 6 5 8 )

0 .7 8 2 �� (0 .0 6 6 2 )

0 .7 7 7 5 �� (0 .0 6 9 6 )

0 .7 8 4 �� (0 .0 7 0 0 )

0 .7 9 7 �� (0 .0 7 3 3 )

H o m e o w n e rs h ip

O w n e rs h ip

o f th e

re si d e n c e

0 .2 0 1 �� (0 .0 6 3 6 )

0 .1 9 7 �� (0 .0 6 3 6 )

0 .1 4 8 �� (0 .0 6 8 1 )

0 .1 4 5 �� (0 .0 6 8 0 )

0 .1 1 0 (0 .0 7 0 0 )

H e a lt h

U n d e rw

e ig h t

¡0 .6 7 7 (0 .4 3 5 )

¡0 .6 3 9 (0 .4 3 4 )

¡0 .6 8 8 (0 .4 4 9 )

¡0 .6 5 0 (0 .4 4 8 )

¡0 .5 7 2 (0 .4 8 1 )

N o rm

a l w e ig h t

R e fe re n c e g ro u p

O v e rw

e ig h t

¡0 .0 5 4 5 (0 .0 6 0 1 )

¡0 .0 3 8 9 (0 .0 6 0 0 )

¡0 .1 0 1 (0 .0 6 3 3 )

¡0 .0 8 6 5 (0 .0 6 3 2 )

¡0 .0 9 1 4 (0 .0 6 4 6 )

O b e si ty

¡0 .1 5 2 �� (0 .0 7 6 5 )

¡0 .1 2 8 � (0 .0 7 6 5 )

¡0 .2 0 6 �� (0 .0 8 1 2 )

¡0 .1 8 5 �� (0 .0 8 1 3 )

¡0 .1 9 9 �� (0 .0 8 3 8 )

D a il y o u td o o r le is u re

a c ti v it ie s

0 .1 9 2 �� (0 .0 6 0 6 )

0 .2 0 8 �� (0 .0 6 0 8 )

0 .1 8 9 �� (0 .0 6 4 3 )

0 .2 0 3 �� (0 .0 6 4 5 )

0 .1 9 5 �� (0 .0 6 6 4 )

P e rs o n a l a tt it u d e s

O w n fi n a n c ia l si tu a ti o n is

v e ry

im p o rt a n t

¡0 .1 8 8 �� (0 .0 5 8 8 )

¡0 .1 9 3 �� (0 .0 5 9 0 )

¡0 .0 1 9 0 �� (0 .0 6 2 7 )

¡0 .1 8 9 �� (0 .0 6 2 9 )

¡0 .1 5 3 �� (0 .0 6 4 7 )

O w n h e a lt h st a tu s is v e ry

im p o rt a n t

0 .1 2 6 �� (0 .0 6 0 8 )

0 .1 2 1 �� (0 .0 6 0 9 )

0 .1 6 3 �� (0 .0 6 4 9 )

0 .1 5 8 �� (0 .0 6 4 9 )

0 .1 5 3 �� (0 .0 6 6 7 )

P ro te c ti o n o f n a tu re

a n d

th e e n v ir o n m e n t is

v e ry

im p o rt a n t

0 .0 4 1 9 (0 .0 5 5 7 )

0 .0 4 2 5 (0 .0 5 5 8 )

0 .0 5 2 2 (0 .0 5 9 6 )

0 .0 5 1 9 (0 .0 5 9 6 )

¡0 .0 3 5 4 (0 .0 6 1 2 )

S e c u ri ty

fr o m

c ri m e s is

v e ry

im p o rt a n t

0 .0 5 6 9 (0 .0 5 7 0 )

0 .0 6 4 8 (0 .0 5 7 4 )

0 .0 2 2 7 (0 .0 6 0 8 )

0 .0 2 6 1 (0 .0 6 1 2 )

¡0 .0 1 4 0 (0 .0 6 3 1 )

S ta te d g e n e ra l ti m e

p re fe re n c e (h ig h v a lu e s:

h ig h p a ti e n c e )

0 .8 4 2 �� (0 .0 1 2 2 )

.0 8 4 1 �� (0 .1 2 2 )

0 .0 7 9 0 �� (0 .0 1 3 1 )

0 .0 7 9 9 �� (0 .0 1 3 1 )

0 .0 8 2 9 �� (0 .0 1 3 5 )

S ta te d g e n e ra l w il li n g n e ss

to ta k e ri sk s

0 .1 6 0 �� (0 .0 1 5 2 )

0 .1 6 2 �� (0 .0 1 5 3 )

0 .1 5 2 �� (0 .0 1 6 4 )

0 .1 5 3 �� (0 .0 1 6 4 )

0 .1 4 9 �� (0 .0 1 6 8 )

S ta te d w il li n g n e ss

to ta k e

ri sk s re g a rd in g o w n

h e a lt h

¡0 .0 7 2 7 �� (0 .0 1 3 5 )

¡0 .0 7 3 2 �� (0 .0 1 3 6 )

¡0 .0 7 0 4 �� (0 .0 1 4 7 )

¡0 .0 7 0 9 �� (0 .0 1 4 7 )

¡0 .0 7 6 4 �� (0 .0 1 5 2 )

(c o n ti n u e d )

2222 D. Osberghaus and J. K€uhling

T a b le 1 .3 .

(C o n ti n u e d )

V a ri a b le

in th e m o d e l

C o e ffi c ie n ts (r o b u st st a n d a rd

e rr o rs )

V a ri a b le in

th e d a ta

C o n tr o ls o n ly

M o d e l 1

M o d e l 2

M o d e l 3

M o d e l 4

F e d e ra l st a te

1 5 D u m m y v a ri a b le s,

re fe re n c e g ro u p :

B a v a ri a

In c lu d e d

D a m a g e

e x p e ri e n c e

(p a st )

P a rt ic ip a n t h a s a lr e a d y e x p e ri e n c e d fi n a n c ia l o r h e a lt h d a m a g e b y …

H e a t w a v e

� ¡0

.6 6 2 �� (0 .1 5 9 )

� ¡0

.5 5 9 �� (0 .1 6 7 )

¡0 .5 8 1 �� (0 .1 7 1 )

S to rm

� ¡0

.0 4 7 5 (0 .0 6 5 5 )

� ¡0

.0 5 0 9 (0 .0 6 8 2 )

¡0 .0 6 3 7 (0 .0 7 0 2 )

H e a v y ra in

� ¡0

.0 2 2 8 (0 .0 6 2 8 )

� ¡0

.0 2 1 1 (0 .0 6 5 4 )

0 .0 1 2 3 (0 .0 6 7 3 )

F lo o d

� ¡0

.1 6 3 � (0 . 8 5 5 )

� ¡0

.1 2 4 (0 .0 8 9 9 )

¡0 .1 5 4 (0 .0 9 3 1 )

C li m a te c h a n g e

e x p e c ta ti o n

(f u tu re )

E x p e c te d c o n se q u e n c e s o f c li m a te c h a n g e o n in d iv id u a l li v in g c o n d it io n s in

th e n e x t d e c a d e s

V e ry

p o si ti v e

� �

0 .4 8 1 (0 .8 8 8 )

0 .4 4 6 (0 .8 9 1 )

0 .4 2 7 (0 .9 0 1 )

R a th e r p o si ti v e

� �

¡0 .0 2 5 3 (0 .1 7 7 )

¡0 .0 5 2 3 (0 .1 7 6 )

¡0 .0 2 7 7 (0 .1 8 2 )

N e it h e r p o si ti v e n o r

n e g a ti v e

R e fe re n c e g ro u p

R a th e r n e g a ti v e

� �

¡0 .1 9 2 �� (0 .0 5 7 9 )

¡0 .2 0 4 �� (0 .0 5 7 9 )

¡0 .2 1 8 �� (0 .0 5 9 6 )

V e ry

n e g a ti v e

� �

¡0 .5 0 9 �� (0 .1 4 4 )

¡0 .5 0 3 �� (0 .1 4 4 )

¡0 .4 6 9 �� (0 .1 5 0 )

C o n st a n t

1 .5 5 3 �� (0 .6 1 9 )

1 .5 6 7 �� (0 .6 2 4 )

1 .8 3 7 �� (0 .6 5 6 )

1 .8 2 1 �� (0 .6 6 2 )

1 .8 3 8 �� (0 .6 8 8 )

O b se rv a ti o n s

4 ,8 2 6

4 ,7 6 6

4 ,2 2 3

4 ,1 7 2

3 ,9 5 4

R 2

0 .1 4 1

0 .1 4 8

0 .1 5 4

0 .1 5 2

N o te : R o b u st st a n d a rd

e rr o rs in

p a re n th e se s. T h e a st e ri sk s (� , �� , a n d �� ) d e n o te si g n ifi c a n c e le v e ls o f 1 0 % , 5 % , a n d 1 % , re sp e c ti v e ly .

Journal of Environmental Planning and Management 2223

T a b le

1 .4 .

O L S re g re ss io n re su lt s. D e p e n d e n t v a ri a b le : li fe

sa ti sf a c ti o n (L S ). S a m p le

fi x e d to

th e m o d e l w it h th e lo w e st n u m b e r o f o b se rv a ti o n s (M

o d e l 3 , N D

4 ,1 7 2 ).

V a ri a b le in

th e m o d e l

V a ri a b le in

th e d a ta

C o e ffi c ie n ts (r o b u st st a n d a rd

e rr o rs )

C o n tr o ls o n ly

M o d e l 1

M o d e l 2

M o d e l 3

A g e

¡0 .0 9 1 6 �� (0 .0 1 8 0 )

¡0 .0 8 7 8 �� (0 .0 1 7 9 )

¡0 .0 9 0 0 �� (0 .0 1 7 9 )

¡0 .0 8 6 5 �� (0 .0 1 7 9 )

A g e 2

0 .0 0 0 9 3 1 �� (0 .0 0 0 1 8 7 )

0 .0 0 0 8 8 7 �� (0 .0 0 0 1 8 7 )

0 .0 0 0 9 1 1 �� (0 .0 0 0 1 8 6 )

0 .0 0 0 8 7 1 �� (0 .0 0 0 1 8 6 )

S e x

S e x (1

D fe m a le )

0 .2 2 5 �� (0 .0 6 9 3 )

0 .2 2 9 �� (0 .0 6 9 2 )

0 .2 2 7 �� (0 .0 6 9 2 )

0 .2 3 0 �� (0 .0 6 9 1 )

F a m il y st a tu s

M a rr ie d , li v in g to g e th e r

0 .0 5 5 6 (0 .0 9 9 8 )

0 .0 4 7 0 (0 .0 9 9 5 )

0 .0 5 7 5 (0 .0 9 9 3 )

0 .0 4 9 2 (0 .0 9 9 0 )

M a rr ie d , li v in g se p a ra te ly

0 .0 6 5 9 (0 .1 6 8 )

0 .0 7 8 8 (0 .1 6 7 )

0 .0 3 7 5 (0 .1 6 8 )

0 .0 5 0 5 (0 .1 6 8 )

S in g le

R e fe re n c e g ro u p

D iv o rc e d

¡0 .1 3 0 (0 .1 2 0 )

¡0 .1 2 6 (0 .1 2 0 )

¡0 .1 2 9 (0 .1 2 0 )

¡0 .1 2 6 (0 .1 2 0 )

W id o w e d

¡0 .2 9 2 (0 .1 9 2 )

¡0 .2 6 2 (0 .1 9 2 )

¡0 .3 2 7 � (0 .1 9 2 )

¡0 .2 9 6 (0 .1 9 2 )

S u c c e ss o rs

C h il d re n (1

D Y e s)

0 .0 3 4 4 (0 .0 8 1 8 )

0 .0 4 1 7 (0 .0 8 1 7 )

0 .0 3 2 2 (0 .0 8 1 7 )

0 .0 3 9 1 (0 .0 8 1 7 )

G ra n d c h il d re n (1

D Y e s)

0 .0 9 0 1 (0 .0 8 2 7 )

0 .0 9 5 7 (0 .0 8 2 8 )

0 .0 8 4 5 (0 .0 8 2 2 )

0 .0 8 9 5 (0 .0 8 2 3 )

E d u c a ti o n

G ra d u a te d fr o m

“ H a u p ts c h u le ”

o r n o t g ra d u a te d

0 .0 7 3 2 (0 .0 9 1 0 )

0 .0 6 5 4 (0 .0 9 0 6 )

0 .0 7 2 5 (0 .0 9 0 8 )

0 .0 6 5 2 (0 .0 9 0 4 )

G ra d . fr o m

“ R e a ls c h u le ” o r re st

R e fe re n c e g ro u p

G ra d . fr o m

h ig h sc h o o l o r

u n iv e rs it y

0 .1 0 7 � (0 .0 6 4 9 )

0 .1 1 2 � (0 .0 6 4 8 )

0 .0 9 8 7 (0 .0 6 4 9 )

0 .1 0 5 (0 .0 6 4 9 )

O c c u p a ti o n

F u ll -t im

e e m p lo y e d

R e fe re n c e g ro u p

P a rt -t im

e e m p lo y e d

0 .0 4 2 4 (0 .0 9 3 7 )

0 .0 5 3 3 (0 .0 9 3 7 )

0 .0 5 6 4 (0 .0 9 3 8 )

0 .0 6 6 5 (0 .0 9 3 8 )

R e ti re d

0 .2 9 8 �� (0 .1 1 0 )

0 .3 1 2 �� (0 .1 0 9 )

0 .3 0 9 �� (0 .1 1 0 )

0 .3 2 2 �� (0 .1 0 9 )

U n e m p lo y e d , se a rc h in g fo r

e m p lo y m e n t

¡0 .5 1 0 � (0 .2 7 5 )

¡0 .4 8 6 � (0 .2 7 0 )

¡0 .4 7 7 � (0 .2 7 2 )

¡0 .4 5 6 � (0 .2 6 8 )

H o u se w if e /- h u sb a n d

¡0 .0 5 5 2 (0 .4 6 1 )

¡0 .0 7 1 3 (0 .4 6 1 )

¡0 .0 4 2 7 (0 .4 4 7 )

¡0 .0 5 8 2 (0 .4 4 7 )

O th e r u n e m p lo y e d , n o t se a rc h in g

fo r e m p lo y m e n t

0 .1 8 8 (0 .2 4 8 )

0 .2 4 7 (0 .2 4 3 )

0 .1 9 3 (0 .2 4 9 )

0 .2 4 9 (0 .2 4 5 )

In c o m e

L n o f H o u se h o ld

in c o m e in

€ 0 .8 0 3 �� (0 .0 7 0 3 )

0 .8 0 1 �� (0 .0 7 0 3 )

0 .7 8 5 �� (0 .0 7 0 1 )

0 .7 8 4 �� (0 .0 7 0 0 )

H o m e o w n e rs h ip

O w n e rs h ip

o f th e re si d e n c e

0 .1 5 1 �� (0 .0 6 8 4 )

0 .1 6 2 �� (0 .0 6 8 1 )

0 .1 3 4 �� (0 .0 6 8 3 )

0 .1 4 5 �� (0 .0 6 8 0 )

(c o n ti n u e d )

2224 D. Osberghaus and J. K€uhling

T a b le 1 .4 .

(C o n ti n u e d )

V a ri a b le in

th e m o d e l

V a ri a b le in

th e d a ta

C o e ffi c ie n ts (r o b u st st a n d a rd

e rr o rs )

C o n tr o ls o n ly

M o d e l 1

M o d e l 2

M o d e l 3

H e a lt h

U n d e rw

e ig h t

¡0 .6 9 2 (0 .4 5 9 )

¡0 .6 3 8 (0 .4 5 8 )

¡0 .7 0 2 (0 .4 5 0 )

¡0 .6 5 0 (0 .4 4 8 )

N o rm

a l w e ig h t

R e fe re n c e g ro u p

O v e rw

e ig h t

¡0 .1 0 1 (0 .0 6 3 3 )

¡0 .0 9 1 0 (0 .0 6 3 2 )

¡0 .0 9 6 0 (0 .0 6 3 3 )

¡0 .0 8 6 5 (0 .0 6 3 2 )

O b e si ty

¡0 .2 2 0 �� (0 .0 8 2 1 )

¡0 .1 9 3 �� (0 .0 8 1 8 )

¡0 .2 0 9 �� (0 .0 8 1 5 )

¡0 .1 8 5 �� (0 .0 8 1 3 )

D a il y o u td o o r le is u re

a c ti v it ie s

0 .1 9 4 �� (0 .0 6 4 6 )

0 .2 0 2 �� (0 .0 6 4 4 )

0 .1 9 6 �� (0 .0 6 4 6 )

0 .2 0 3 �� (0 .0 6 4 5 )

P e rs o n a l a tt it u d e s

O w n fi n a n c ia l si tu a ti o n is v e ry

im p o rt a n t

¡0 .1 9 1 �� (0 .0 6 3 2 )

¡0 .1 9 2 �� (0 .0 6 3 0 )

¡0 .1 8 8 �� (0 .0 6 3 1 )

¡0 .1 8 9 �� (0 .0 6 2 9 )

O w n h e a lt h st a tu s is v e ry

im p o rt a n t

0 .1 5 0 �� (0 .0 6 5 3 )

0 .1 4 7 �� (0 .0 6 5 1 )

0 .1 6 1 �� (0 .0 6 5 1 )

0 .1 5 8 �� (0 .0 6 4 9 )

P ro te c ti o n o f n a tu re

a n d th e

e n v ir o n m e n t is v e ry

im p o rt a n t

0 .0 0 4 7 1 (0 .0 5 9 2 )

0 .0 0 9 3 2 (0 .0 5 9 0 )

0 .0 4 9 0 (0 .0 5 9 7 )

0 .0 5 1 9 (0 .0 5 9 6 )

S e c u ri ty

fr o m

c ri m e s is v e ry

im p o rt a n t

0 .0 3 2 3 (0 .0 6 1 4 )

0 .0 3 2 5 (0 .0 6 1 4 )

0 .0 2 6 1 (0 .0 6 1 2 )

S ta te d g e n e ra l ti m e p re fe re n c e

(h ig h v a lu e s: h ig h p a ti e n c e )

0 .0 8 3 7 �� (0 .0 1 3 2 )

0 .0 8 1 9 �� (0 .0 1 3 1 )

0 .0 8 1 6 �� (0 .0 1 3 1 )

0 .0 7 9 9 �� (0 .0 1 3 1 )

S ta te d g e n e ra l w il li n g n e ss

to ta k e

ri sk s

0 .1 5 5 �� (0 .0 1 6 5 )

0 .1 5 7 �� (0 .0 1 6 5 )

0 .1 5 1 �� (0 .0 1 6 5 )

0 .1 5 3 �� (0 .0 1 6 4 )

S ta te d w il li n g n e ss

to ta k e ri sk s

re g a rd in g o w n h e a lt h

¡0 .0 7 1 2 �� (0 .0 1 4 8 )

¡0 .0 7 2 0 �� (0 .0 1 4 7 )

¡0 .0 7 0 1 �� (0 .0 1 4 8 )

¡0 .0 7 0 9 �� (0 .0 1 4 7 )

F e d e ra l st a te

1 5 D u m m y v a ri a b le s, re fe re n c e

g ro u p : B a v a ri a

In c lu d e d

D a m a g e e x p e ri e n c e

(p a st )

P a rt ic ip a n t h a s a lr e a d y e x p e ri e n c e d fi n a n c ia l o r h e a lt h d a m a g e b y …

H e a t w a v e

� ¡0

.5 9 1 �� (0 .1 6 8 )

� ¡0

.5 5 9 �� (0 .1 6 7 )

S to rm

� ¡0

.0 5 7 3 (0 .0 6 8 4 )

� ¡0

.0 5 0 9 (0 .0 6 8 2 )

H e a v y ra in

� ¡0

.0 2 2 7 (0 .0 6 5 5 )

� ¡0

.0 2 1 1 (0 .0 6 5 4 )

F lo o d

� ¡0

.1 2 5 (0 .0 9 0 3 )

� ¡0

.1 2 4 (0 .0 8 9 9 )

(c o n ti n u e d )

Journal of Environmental Planning and Management 2225

T a b le 1 .4 .

(C o n ti n u e d )

V a ri a b le in

th e m o d e l

V a ri a b le in

th e d a ta

C o e ffi c ie n ts (r o b u st st a n d a rd

e rr o rs )

C o n tr o ls o n ly

M o d e l 1

M o d e l 2

M o d e l 3

C li m a te c h a n g e

e x p e c ta ti o n (f u tu re )

E x p e c te d c o n se q u e n c e s o f c li m a te c h a n g e o n in d iv id u a l li v in g c o n d it io n s in

th e n e x t d e c a d e s

V e ry

p o si ti v e

� �

0 .4 4 6 (0 .8 8 4 )

0 .4 4 6 (0 .8 9 1 )

R a th e r p o si ti v e

� �

¡0 .0 4 0 5 (0 .1 7 7 )

¡0 .0 5 2 3 (0 .1 7 6 )

N e it h e r p o si ti v e n o r n e g a ti v e

R e fe re n c e g ro u p

R a th e r n e g a ti v e

� �

¡0 .2 0 5 �� (0 .0 5 8 0 )

¡0 .2 0 4 �� (0 .0 5 7 9 )

V e ry

n e g a ti v e

� �

¡0 .5 3 7 �� (0 .1 4 5 )

¡0 .5 0 3 �� (0 .1 4 4 )

C o n st a n t

1 .6 0 6 �� (0 .6 6 4 )

1 .5 8 2 �� (0 .6 6 3 )

1 .8 5 2 �� (0 .6 6 2 )

1 .8 2 1 �� (0 .6 6 2 )

O b se rv a ti o n s

4 ,1 7 2

R 2

0 .1 4 5

0 .1 4 9

0 .1 5 0

0 .1 5 4

2226 D. Osberghaus and J. K€uhling

Table 1.5. Questions and answer options of the key variables LS, damage experience, and damage expectations (translated from German). The “don’t know” option was possible in each question.

Variable in the data Question Options

Self-rated life satisfaction In general, how satisfied are you currently with your life?

Eleven categories, of which the lowest is named “totally dissatisfied” and the highest “totally satisfied”

Damage experience from extreme weather events

In the following various natural events are listed. Please mark each which you have personally experienced at home, at work or during a journey. If one or more of the events have been marked, the marked events have been presented again with this follow-up question: Please mark now for each event, whether you have suffered any financial or health damage (with consultation of a doctor) from the event.

� Heat waves (e.g. such that you did not want to be outside and changed your plans accordingly)

� Storms (e.g. such that you have avoided leaving your home)

� Heavy rain or hail (e.g. such that you have worried about your car, garden or house)

� Floods or inundation

Expected consequences of climate change on individual living conditions in the next decades

According to your assessment, which consequences will climate change have for your very personal living conditions in the next decades?

� Very positive consequences � Rather positive consequences � Broadly equally negative and

positive consequences � Rather negative consequences � Very negative consequences

Journal of Environmental Planning and Management 2227

Table 1.6. Regression results (coefficients and robust standard errors) of simultaneous estimations by the Stata command “cmp” (Roodman 2011).

Variable in the model Variable in the data

Dependent variable: life satisfaction

(OLS)

Dependent variable: damage

expectations (ordered probit)

Age Age ¡0.0866�� (0.0178) ¡0.0123 (0.0114) Age

2 0.000873

�� (0.000185) 0.0000927 (0.000115)

Sex Sex (1 D female) 0.228�� (0.0689) 0.0247 (0.0464) Family status Married, living together 0.0490 (0.0985) 0.0404 (0.0655)

Married, living separately 0.0558 (0.167) ¡0.164 (0.111) Single Reference group

Divorced ¡0.125 (0.119) ¡0.0171 (0.0775) Widowed ¡0.288 (0.192) ¡0.254�� (0.117)

Successors Children (1 D Yes) 0.0393 (0.0811) ¡0.0267 (0.0545) Grandchildren (1 D Yes) 0.0911 (0.0818) ¡0.0315 (0.0551)

Education Graduated from “Hauptschule” or not graduated

0.0643 (0.0900) 0.0322 (0.0590)

Grad. from “Realschule” or rest

Reference group

Grad. from high school or university

0.106 � (0.0645) ¡0.0652 (0.0429)

Occupation Full-time employed Reference group

Part-time employed 0.0638 (0.0934) 0.0759 (0.0603)

Retired 0.320 ��

(0.109) 0.0599 (0.0689)

Unemployed, searching for employment

-0.463 � (0.267) 0.206 (0.156)

Housewife/-husband ¡0.0612 (0.447) 0.0698 (0.260) Other unemployed, not

searching for employment 0.249 (0.243) ¡0.0350 (0.142)

Income Ln of Household income in € 0.788 ��

(0.0702) ¡0.170�� (0.0442) Homeownership Ownership of the residence 0.148

�� (0.0682) ¡0.0855� (0.0441)

Health Underweight ¡0.649 (0.448) ¡0.0344 (0.233) Normal weight Reference group

Overweight ¡0.0881 (0.0630) 0.0865�� (0.0422) Obesity ¡0.186�� (0.0810) 0.0829 (0.0532) Daily outdoor leisure

activities 0.203

�� (0.0641) ¡0.0261 (0.0420)

Personal attitudes

Own financial situation is very important

¡0.190�� (0.0625) 0.0228 (0.0414)

Own health status is very important

0.156 �� (0.0648) 0.0737

� (0.0436)

Protection of nature and the environment is very important

0.0424 (0.0631) 0.108 ��

(0.0417)

Security from crimes is very important

0.0280 (0.0611) ¡0.00745 (0.0414)

0.0804 ��

(0.0131) ¡0.0139� (0.00805) (continued)

2228 D. Osberghaus and J. K€uhling

Table 1.6. (Continued )

Variable in the model Variable in the data

Dependent variable: life satisfaction

(OLS)

Dependent variable: damage

expectations (ordered probit)

Stated general time preference (high values: high patience)

Stated general willingness to take risks

0.154 ��

(0.0165) ¡0.0298�� (0.00975)

Stated willingness to take risks regarding own health

¡0.0711�� (0.0147) 0.00871 (0.00901)

Combatting climate change is very important

� 0.510�� (0.0419)

Partisanship of a left wing party

� 0.103�� (0.0388)

Agreement with statement “Humans are mainly responsible for climate change”

� 0.398�� (0.0393)

Agreement with building of new coal power plants

� ¡0.132�� (0.0515)

Information source for daily news: Internet

� ¡0.0797�� (0.0375)

Federal state 15 Dummy variables, reference group: Bavaria

Included

Damage experience (past)

Participant has already experienced financial or health damage by…

Heat wave ¡0.567�� (0.167) 0.290�� (0.101) Storm ¡0.0524 (0.0679) 0.0562 (0.0465) Heavy rain ¡0.0214 (0.0650) 0.0222 (0.0451) Flood ¡0.123 (0.0894) ¡0.0418 (0.0622)

Climate change expectation (future)

Expected consequences of climate change on individual living conditions in the next decades

Very positive 0.345 (0.911) � Rather positive ¡0.108 (0.217) Neither positive nor negative Reference group

Rather negative ¡0.158 (0.116) Very negative ¡0.409 (0.251)

Constant 1.765 ��

(0.670) ¡ Threshold 1 (v1sim) � ¡4.612 (0.457) Threshold 2 (v2sim) ¡3.448 (0.434) Threshold 3 (v3sim) ¡1.358 (0.430) Threshold 4 (v4sim) 0.382 (0.429)

Rho (measure of correlation between error terms)

¡0.0214 (0.0483)

Observations 4174

Note: Robust standard errors in parentheses. The asterisks ( � , �� , and

�� ) denote significance levels of 10%, 5%,

and 1%, respectively.

Journal of Environmental Planning and Management 2229

Table 1.7. Calculation of the indirect effect of damage experience on LS, using simultaneous estimations by the Stata command “cmp” (Roodman 2011).

Climate damage expectation levels (Ci)

Average marginal effect (change of estimated probability with regard to heat wave experience)

LS effect of expectation level

Indirect effects

Ci D 1 ¡0.00121� (0.000637) 0.345 (0.911) ¡0.000416 (0.00110) Ci D 2 ¡0.0154�� (0.00555) ¡0.108 (0.217) 0.00167 (0.00334) Ci D 3 ¡0.0879�� (0.0304) 0 (reference group) 0 Ci D 4 0.0726�� (0.0251) ¡0.158 (0.116) ¡0.0115 (0.00845) Ci D 5 0.0320�� (0.0112) ¡0.409 (0.251) ¡0.0131 (0.00801) Sum over expectation

levels (λsim) n.a. n.a. ¡0.0232� (0.0122)

Note: The asterisks ( � , �� , and

�� ) denote significance levels of 10%, 5%, and 1%, respectively. Robust standard

errors of marginal effects and coefficients in parentheses. Standard errors in column 4 have been calculated manually using the error propagation formulas given in Taylor (1997).

2230 D. Osberghaus and J. K€uhling

Copyright of Journal of Environmental Planning & Management is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

Abstract
1. Introduction and literature
2. Theoretical model
3. Data

3.1. Subjective well-being
3.2. Damage experience
3.3. Climate change expectation

4. Empirical strategy
5. Results
6. Discussion of results

6.1. Damage experience and life satisfaction
6.2. Damage expectations and life satisfaction
6.3. Direct and indirect LS-Effects of damage experience
6.4. Estimation of climate damage expectations

7. Conclusions
Acknowledgements
Funding
References