Content uploaded by Shafin Shabir
Author content
All content in this area was uploaded by Shafin Shabir on Sep 21, 2018
Content may be subject to copyright.
Hard Times and Religion: Do Religious People Stay
Happier in Times of Crisis?
Shafin Shabir ∗
30 June 2018
Numerous studies have concluded that religious people are more likely to
be happier in life. In this paper I ask whether people who consider themselves
to be spiritual are more likely to report higher subjective life satisfaction in
times of an economic crisis than those who are less spiritual. Using data
from European Social Survey (Rounds 1-8), I find that on an average people
who are more spiritual have a higher life satisfaction in times of crisis than
people who report lower level of spirituality.
1 Introduction
Religion is probably as old as the human civilisation. One could easily
say that religion to this day remains a very strong institution, withstanding
all the ’atrocities’ that it committed or that it had to face. My aim here is
not do discuss or analyse the institutional part of the religion. Rather I am
more interested in its spiritual aspects and the way it influences a person’s
subjective well-being.
In the post-modern era, which happens to be my period of interest, the
idea of religion and spirituality as different ideas has gained popularity.1
A lot of people consider themselves spiritual but at the same time do not
want to be associated with a particular religion. Approximately 30% of peo-
ple who said that they do not identify themselves to a particular religious
denomination consider themselves to be spiritual.2This disinclination to-
wards an institutionalized form if religion may be due to various reasons
∗Barcelona Graduate School of Economics (shafin.shabir@barceloonagse.eu)
1For an overview see: https://en.wikipedia.org/wiki/Spiritual_but_not_religious#
cite_note-FOOTNOTEMercandante2014-1
2This has been taken from the ESS data(2002-2016) where being spiritual is measured on a
scale from 0-10 with 0 being not at all spiritual to 10 being the most spiritual. For the purpose
of this statistic we coded spiritual with a score ≥5.
1
some of them are mentioned in a book titled Spiritual but not religious: a
call to religious revolution in America written by Erlandson (2000).
But why do people become spiritual despite not wanting to follow a re-
ligion? One of the reasons might be the positive impact of spiritulity on
life satisfaction (For a review of the literature see Swinyard et al. (2001)).
Ferriss (2002) also finds a positive correlation between religiousness and the
quality of life and a negative impact with the indicators of stress. Taking
this as my motivation, I am interested to know whether spiritual people
remain happier in times of crisis than those who are not spiritual. I use
the European crises as an external shock to analyse the impact of spiritual
people on their life satisfaction. My results suggest that people who con-
sider themselves as spiritual remain happier than others even in times of
crisis.
In the following section I give a brief introduction of the literature on
the topic. In section 3, I discuss the datasets that I have used in the paper
which is followed by empirical specification and results in section 4. I end
with a conclusion in section 5.
2 Literature Review
A lot of work has been done in the subjective measure of well being.
Usually the surveys designed to measure well being ask the question: How
satisfied are you with life? and the respondent has to choose on a scale of
0-103where 0 means not at all satisfied and 1 means completely satisfied.
The basic assumptions that we need to make when doing an analysis using
this variable are that the measures can be compared and individuals can
value their satisfaction on a numeric scale. The assumption of ordinality
or cardinality is not required in such an analyses(Ferrer-I-Carbonell and
Frijters, 2004). Fleurbaey and Schwandt (2015) find that a significant
portion of people try to maximize their subjective well-being. However,
they also discover the ’paradox of happiness’- those who try to maximize
their subjective well being have a lower level of satisfaction then those who
maximize some other goals. One of the reason for this paradox might be
due to the inability of these people to actually reach a maximum level
3It can also be from 0-7 or any other number depending on the questionnaire
2
of satisfaction while the latter by selecting objective goals might become
comparatively happier on achieving such a goal.
The literature on the relationship between religion and subjective well
being is also abundant. We have already mentioned some texts in the In-
troduction of this paper which give a brief review of the entire literature on
religion and subjective well-being. Additionally, Helliwell (2003) using the
World Values Survey finds a positive spillover effects of weekly church at-
tendance and the well being of others at the national level. This essentially
can be viewed as a positive externality of being religious.4
The fact that spiritual people are more satisfied in life, raises an interest-
ing question, which to the knowledge of the author has not been attempted
to answer- Is the satisfaction of spiritual people ’robust’ to shocks in the
external environment, i.e., is the subjective well being of these people not
as much affected by the shocks as those who consider themselves as being
non-spiritual.
3 Data
I am using data from the European Social Survey for all the waves be-
tween 2002 to 2016. The ESS data is collected with two year intervals
which means we have eight rounds of the survey results. I merged all
the data and our required variables to create a big dataset. I only kept the
variables which very available for all the rounds and were related our depen-
dent variable (life satisfaction). This list of variables includes religiousness
(spirituality), marital status, education (self, father and mother), age and
gender.
Subjective well-being or life satisfaction is measured using the variable
called happiness. In the survey the following question is asked: Taking
all things together, how happy would you say you are?. The respondents
are asked to choose on a scale from zero to 10 where the former means
”Extemely unhappy” and the latter means ”Extremely Happy”. Figure 1
summaries the variable for the waves of the ESS data. The distribution
4The estimates are not causal
3
in 2002 is more titled towards the happy side and in the consecutive years
remains similar (more or less).
Figure 1: Figure 1
Our another variable of interest is the variable which measures spiritu-
ality. The question asked to the respondent is: Regardless of whether you
belong to a particular religion, how religious would you say you are?. The
respondents are asked to choose on a scale from 0 to 10 where 0 means
”Not at all religious” and 10 means ”Very Religious”. Since this question
is asked to people regardless of whether they follow a particular religion
or not, I have taken this variable to mean as the self reported measure of
spirituality. Table 1 gives a summary statistic of this variable from 2002-
2016. It reports the percentage distribution of people within the categories
of this variable for a particular religion. For example, in 2004 around 7%
of the people reported that they are very religious while 12% of the people
said that they are very religious.
Summary statistics of education(self, father and mother) and legal mari-
tal status is provided in the appendix of this paper. I have harmonized the
data5as per the first round (2002) and use statistical weights in reporting
all the Summary Tables and also in the regression.
5For example marital status had only 5 categories in 2002 but in 2008 it had 8 categories.
So I bought all the other years to the same level as in 2002. I did a similar coding for the
education variable.
4
Table 1: Summary Statistics- How Religious Are You?
year
2002 2004 2006 2008 2010 2012 2014 2016 Total
% % % % % % % % %
Not at all religious 13.1 11.9 14.6 12.7 14.2 14.5 19.0 17.4 14.5
1 5.0 4.9 5.8 5.4 5.7 5.4 5.9 5.7 5.5
2 7.1 6.5 7.9 6.8 7.6 7.1 7.2 7.6 7.2
3 8.1 7.8 8.6 8.2 8.9 8.2 7.9 8.8 8.3
4 6.6 6.8 7.2 6.6 6.9 6.5 6.7 6.5 6.7
5 17.3 18.3 18.6 17.9 18.6 17.0 14.9 15.8 17.4
6 10.7 10.2 9.2 9.7 9.8 10.4 9.3 9.8 9.9
7 11.8 11.4 10.5 11.8 11.4 10.9 10.4 10.7 11.2
8 10.1 10.4 9.1 10.5 8.8 10.1 9.1 8.9 9.7
9 4.4 4.8 3.4 4.3 3.7 4.0 3.8 3.7 4.0
Very religious 5.8 6.8 5.0 6.0 4.3 6.0 5.7 4.9 5.6
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
4 Empirical Specifications and Discussions
4.1 Regressions
My analysis relies on the following specifications:
happyijt =β0+α1crisisyeart+α2rlgdgrij t +λcrisisyeartrldgrijt +β1Xit +it
(1)
where
•happyijt = Measures the subjective well-being of individual i in coun-
try j at year t
•crisisyeart= Is a dummy for the crisis year (2010 and 2012 waves
corresponding to the crisis)
•rlgdgrij t = Measures the spirituality of individual i in country j at
year t
•X0
ijt controls for education (self, father and mother), age, gender and
legal marital status
In the given equation our variable of interest is λ. We can interpret the
coefficient in the above given equation as an average difference between
life satisfaction and happiness levels of those with marginally higher level
of spirituality in a crisis year against the marginal happiness of spiritual
people in non-crisis year.
5
Despite controlling for all the stuff, using a panel data allows us to con-
trol for more heterogeneity. We can control for country fixed and year fixed
effects. This will control for all the time invariant region specific or time
specific effects and therefore our estimates will be closer to the ’real esti-
mates’ for the population. We do this in the following specification where ηt
and ηjcontrol for the time fixed and the country fixed effects respectively.
happyijt =β0+ηt+ηj+α1rlgdgrijt +λcrisisyeartrldgrijt +β1Xit +it (2)
Before presenting the results a note of caution. The estimates from these
equation cannot be treated as causal in nature. The equation even after
controlling for fixed effects suffers from the problems of omitted variable
bias. It might be the case people’s personal behaviour is such that they
are less prone to external shocks and therefore might stay happier even in
times of crisis. Since we cannot control for individual types, this violates
the assumption of the exogeniety of the error terms. We will see this
effect in our regression equation via our variable of interest. Therefore, the
estimates from our regression will be biased upwards.
4.2 Results and Discussion
The results from our specification are shown in Table 2. On an average
people report lower satisfaction levels in the crisis year which is intuitive
as people are negatively affected by the crisis which should reflect on their
subjective well-being. I also find that on an average spiritual people are
more happier than the less spiritual people. It is important to mention
here that we are using a normal OLS and not ordinal logit as in the case of
subjective well-being both provide similar results (Ferrer-I-Carbonell and
Frijters, 2004).
As expected, λhas a positive sign and the coefficient is statistically
significant. The coefficient is probably biased upwards as argued in the
previous subsection. Moreover, adding country fixed effects and year fixed
effects decreases the magnitude of the coefficient which is further supports
the idea that the value is biased upwards.
But why does a positive coefficient make sense? There are a couple of
reasons that I can think of. Firstly, people might have started becoming
6
Table 2: Results
(1) (2)
happy happy
Subjective Well Being
Crisis Year -0.115∗∗∗
(0.0278)
Spirituality 0.0272∗∗∗ 0.0602∗∗∗
(0.00275) (0.00274)
Crisis Year*Sprituality 0.0248∗∗∗ 0.0108∗
(0.00508) (0.00494)
Gender -0.0296∗-0.0409∗∗
(0.0137) (0.0129)
Calculated Age -0.00290∗∗∗ -0.0109∗∗∗
(0.000519) (0.000504)
Legal Marital Status
Married (Base Category)
Separated -0.968∗∗∗ -1.052∗∗∗
(0.0650) (0.0619)
Divorced -0.830∗∗∗ -0.815∗∗∗
(0.0273) (0.0256)
Widowed -1.086∗∗∗ -0.784∗∗∗
(0.0302) (0.0289)
Never Married -0.130∗∗∗ -0.375∗∗∗
(0.0186) (0.0178)
Highest Level of Education
Less than lower secondary education (Base Category)
Lower secondary education completed 0.129∗∗∗ 0.0643∗
(0.0288) (0.0277)
Upper secondary education completed -0.115∗∗∗ 0.00602
(0.0197) (0.0189)
Post-secondary non-tertiary education 0.255∗∗∗ 0.0533∗
(0.0284) (0.0269)
Tertiary education completed -0.155∗∗∗ 0.106∗∗∗
(0.0290) (0.0274)
Father’s Education YES YES
Mother’s Education YES YES
Year Fixed Effects No YES
Country Fixed Effects No YES
Number of Observations 285153 285153
R20.041 0.155
Robust standard errors in parentheses
∗p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001 7
more spiritual after a shock in their personal life history to protect their
subjective well being in the future from such external shocks. In other
words, people might have self selected themselves to becoming spiritual in
order to escape the hardships of this world. Therefore, the people who
believe that they are more spiritual will by construction be happier.
Secondly, the entire philosophy behind spirituality is looking inwards and
protecting yourself from external desires. Therefore, an external shock like
an Economic Crisis will have lesser impact on these people compared to
others who are not spiritual.
More education is positively related to happiness. But this sign and
estimate can also be because of the income a person. It can be the case that
more educated people are more likely to be employed and therefore might
earn higher income. And, according to a study conducted by Frijters et al.
(2004) income and life satisfaction are positively correlated. Therefore, the
part of the magnitude of the coefficient that we see in Table 2 can be via
the income channel.
The base category for legal marital status is ’Married’ as shown in Table
2. Compared to Married, every other category has lower subjective life
satisfaction. This can be because of lesser number of observations for ’Sep-
arated’, ’Divorced’ and ’Widowed’. But what the data suggests that on an
average in Europe, married people generally report higher life satisfaction
as compared to ’Separated’, ’Divorced’ ’Widowed’ and ’Never Married’.
Once again the estimates from the estimation are not causal in nature.
Apart from this, I also do a difference in difference type of estimation
where I consider a person to be spiritual if her reported subjective happiness
is greater than 6. The results from this estimation are given in Table 3.
The results are similar to Table 2. In this case λcan be interpreted in the
following way: the average difference between the subjective life satisfaction
of the spiritual people and the non-spiritual people after the crisis year
minus the average difference between the life satisfaction between spiritual
and non-spiritual before the crisis is positive. This essentially means that
the life satisfaction of spiritual people is higher than non-spiritual people
even during a crisis year.
8
Table 3: Results: Difference in Difference Type of Estimation
(1) (2)
happy happy
Subjective Well Being
Crisis Year -0.0519∗∗ 0.264∗∗∗
(0.0183) (0.0264)
Spirituality 0.242∗∗∗ 0.363∗∗∗
(0.0168) (0.0163)
Crisis Year*Sprituality 0.164∗∗∗ 0.0792∗
(0.0327) (0.0313)
Gender -0.0308∗-0.0562∗∗∗
(0.0136) (0.0128)
Calculated Age -0.00292∗∗∗ -0.0107∗∗∗
(0.000517) (0.000503)
Legal Marital Status
Married (Base Category)
Separated -0.961∗∗∗ -1.057∗∗∗
(0.0647) (0.0616)
Divorced -0.831∗∗∗ -0.827∗∗∗
(0.0272) (0.0255)
Widowed -1.092∗∗∗ -0.791∗∗∗
(0.0301) (0.0288)
Never Married -0.132∗∗∗ -0.384∗∗∗
(0.0185) (0.0177)
Highest Level of Education
Less than lower secondary education (Base Category)
Lower secondary education completed 0.129∗∗∗ 0.0599∗
(0.0287) (0.0277)
Upper secondary education completed -0.115∗∗∗ 0.00174
(0.0196) (0.0189)
Post-secondary non-tertiary education 0.249∗∗∗ 0.0467
(0.0285) (0.0272)
Tertiary education completed -0.154∗∗∗ 0.105∗∗∗
(0.0291) (0.0277)
Father’s Education YES YES
Mother’s Education YES YES
Year Fixed Effects No YES
Country Fixed Effects No YES
Number of Observations 287329 287329
R20.042 0.154
Robust standard errors in parentheses
∗p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001 9
5 Conclusion
In this paper, I find that spiritual people are more likely to report higher
life satisfaction in times of an economic crisis than the group of people who
do not consider themselves spiritual. I argue that the results, though not
causal, are intuitive and logical. I use a difference in difference type of
approach to answer this problem.
However, in order to correctly identify the problem a survey that would
track the individuals across time would be very useful. In such a survey we
could also control for time invariant individual specific effects which will
bring the estimates closer to the ’actual’ estimate.
10
References
Sven E. Erlandson. Spiritual but not religious: a call to religious revolution
in America. Writers Showcase, 2000.
Ada Ferrer-I-Carbonell and Paul Frijters. How important is methodology
for the estimates of the determinants of happiness?*. The Economic
Journal, 114(497):641–659, 2004.
A L Ferriss. Religion and the quality of life. Journal of Happiness Studies,
3(3):199–215, Sep 2002. URL https://link.springer.com/article/
10.1023/A:1020684404438.
Marc Fleurbaey and Hannes Schwandt. Do people seek to maximize their
subjective well-being –and fail? Working paper IZA DP No. 9450, 2015.
Paul Frijters, John P. Haisken-DeNew, and Michael A. Shields. Money
does matter! evidence from increasing real income and life satisfaction
in east germany following reunification. American Economic Review, 94
(3):730–740, June 2004. doi: 10.1257/0002828041464551. URL http:
//www.aeaweb.org/articles?id=10.1257/0002828041464551.
John F Helliwell. How’s life? combining individual and national vari-
ables to explain subjective well-being. Economic Modelling, 20(2):331 –
360, 2003. ISSN 0264-9993. URL âĂćttp://www.sciencedirect.com/
science/article/pii/S0264999302000573. Henry Special Issue.
European Social Survey. Ess data rounds 1-8. Norwegian Centre for Re-
search Data, Norway, 2002-2016.
W R Swinyard, A K Kau, and H Y Phua. Happiness, materialism, and re-
ligious experience in the us and singapore. Journal of Happiness Studies,
2:13–32, 2001.
11
Appendix
Table 4: Summary Statistics- Highest level of education
year
2002 2004 2006 2008 2010 2012 2014 2016 Total
% % % % % % % % %
Less than lower secondary education (ISCED 0-1) 17.2 21.8 12.1 14.4 6.5 5.9 7.4 5.3 11.4
Lower secondary education completed (ISCED 2) 21.9 18.2 16.3 16.0 15.2 15.7 17.6 12.9 16.6
Upper secondary education completed (ISCED 3) 37.7 35.0 36.6 34.0 37.2 35.0 36.6 33.1 35.6
Post-secondary non-tertiary education completed (ISCED 4) 2.5 2.6 2.5 3.1 3.8 4.0 5.6 4.2 3.4
Tertiary education completed (ISCED 5-6) 20.7 22.4 32.6 32.5 37.4 39.4 32.7 44.5 33.0
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Table 5: Summary Statistics- Father’s Highest level of education?
year
2002 2004 2006 2008 2010 2012 2014 2016 Total
% % % % % % % % %
Less than lower secondary education (ISCED 0-1) 39.9 37.6 30.4 38.0 24.2 25.0 26.5 20.7 30.7
Lower secondary education completed (ISCED 2) 16.8 20.7 21.2 15.6 19.5 19.7 18.5 18.2 18.7
Upper secondary education completed (ISCED 3) 28.9 26.1 26.5 24.5 29.2 27.7 31.6 29.6 27.7
Post-secondary non-tertiary education completed (ISCED 4) 0.9 1.0 1.0 2.2 2.4 2.7 3.3 2.0 1.9
Tertiary education completed (ISCED 5-6) 13.5 14.6 20.8 19.7 24.7 24.9 20.1 29.5 21.0
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
Table 6: Summary Statistics- Mother’s highest level of education?
year
2002 2004 2006 2008 2010 2012 2014 2016 Total
% % % % % % % % %
Less than lower secondary education (ISCED 0-1) 44.7 41.9 32.8 41.1 26.7 27.8 30.0 22.1 33.7
Lower secondary education completed (ISCED 2) 25.4 27.3 26.1 19.0 23.8 22.4 23.6 21.1 23.4
Upper secondary education completed (ISCED 3) 22.5 21.4 22.7 21.6 25.2 25.9 29.5 28.0 24.2
Post-secondary non-tertiary education completed (ISCED 4) 0.9 0.9 0.9 1.8 2.2 2.5 3.2 2.2 1.8
Tertiary education completed (ISCED 5-6) 6.4 8.6 17.5 16.6 22.1 21.4 13.8 26.5 16.9
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
12
Table 7: Summary Statistics- Marital Status
year
2002 2004 2006 2008 2010 2012 2014 2016 Total
% % % % % % % % %
Married 58.6 58.1 57.4 58.0 55.2 55.2 55.9 54.0 56.5
Separated 1.7 1.3 1.3 1.3 0.5 0.5 0.4 0.4 0.9
Divorced 5.2 5.8 7.2 6.9 9.0 8.7 8.3 9.5 7.7
Widowed 7.0 7.4 8.6 8.1 7.8 7.4 6.0 6.9 7.5
Never married 27.6 27.4 25.5 25.7 27.5 28.2 29.3 29.3 27.4
Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
13
