Advertisement

Economics of Disasters and Climate Change

, Volume 2, Issue 3, pp 259–281 | Cite as

Turning to God in Tough Times? Human Versus Material Losses from Climate Disasters in Canada

  • Oscar Zapata
Original Paper

Abstract

Formal and informal insurance mechanisms help people recover from material losses associated with climate disasters. However, people may also find other ways to cope with human losses caused by disasters and research has suggested that religion may provide psychological relief to individuals experiencing adversity. Here, I test whether climate disasters have a causal effect on religious preferences and the intensity of these preferences across provinces in Canada. I look at the differentiated effect of material and human losses on religiosity. I create a dataset with socioeconomic and demographic information of individuals, including their religious preferences, and information on climate disasters at the provincial level in Canada for the period 1992–2012 and use an instrumental variable approach to deal with omitted variables. The novel finding of this paper is that the frequency of disasters and their impacts have different effects on religious preferences: 1) the number and the economic costs of disasters erode religion preferences, and 2) among religious individuals, human losses increase the intensity of their religious preferences. I also find that disasters at the country-wide level influence religious preferences at the local level.

Keywords

Climate disasters Religious preferences Material and human losses IV estimation Canada 

JEL Classification

Q54 Z12 D12 

Noah’s Ark

The waters flooded the earth for a hundred and fifty days (Genesis, 7:24).

Then Noah built an altar to the Lord and, taking some of all the clean animals and clean birds, he sacrificed burnt offerings on it. 21 The Lord smelled the pleasing aroma and said in his heart: “Never again will I curse the ground because of humans, even though every inclination of the human heart is evil from childhood. And never again will I destroy all living creatures, as I have done.

(Genesis, 8:20–21)

Introduction

The increased number and intensity of climate disasters around the world is one of the consequences of climate change. Human and economic losses associated to climate disasters have been increasing over the last three decades and are expected to become even larger worldwide in the near future (Hoeppe 2016). In the context of climate change, where climate disasters are expected to occur more often and with greater intensity, the identification of mechanisms that help people cope with natural disasters, and their economic and human consequences, has become a topic of increasing interest. To some extent insurance markets have helped alleviate the material cost of disasters.1 However many valuable assets in society cannot be covered by insurance and require alternative social or individual coping mechanisms. The psychological costs associated with human losses (i.e., mortality and morbidity), emotional stress, anxiety and fear constitute a burden for people affected directly or indirectly by disasters. Research suggests, for example, that people turn to religion for psychological relief and comfort in periods of adversity and uncertainty (Scheve and Stasavage 2006). Historically, the need of social norms (Wilson 2012; Diamond 2012) and fear to unknown natural phenomena (Stern 2007; Dennett 2006) explain the origin of religious beliefs as early as 33,000 years ago (Kneale 2014). Religiosity, as an economic preference, can be defined not only in terms of a person’s belief that God exists, which in turn may influence individual and collective economic outcomes, but also in terms of practices that involve the use of scarce resources, such as time spent in attending religious services or rituals (Iannaccone 1992, 1995, 1988). In this paper, I study the effect of climate disasters on religious preferences from an empirical perspective. Specifically, I focus on whether climate disasters make people more or less religious and how the intensity of religious preferences changes as a result of these disasters.

I study these relationships in Canada, a country with strong institutions and insurance markets that help mitigate and adapt to the effects of natural disasters. Moreover, Canada is a relatively disaster-safe country in the world, and finding that religion helps in this context would show the importance of religion as a coping mechanism to obtain relief in an environment of climate uncertainty. To study this, I use information about religious preferences at the individual level (i.e., whether the person believes in God, and how often the person attends religious services) from the International Social Survey Program (ISSP). For my analysis, I use the ISSP survey waves from 1992 to 2012, and augment the survey responses with data on the occurrence of climate disasters at the provincial level collected by Public Safety Canada. Information about disasters includes the number of events, human losses (i.e., number of fatalities and injured people) and economic costs.

Technically, omitted variables in the analysis can compromise the identification of the causal effect of climate disasters on religious preferences. Unobserved individual characteristics that correlate with location choices can bias the OLS results. For instance, people’s risk preferences may determine whether they live in a disaster-prone area, or the degree of preparedness to face disasters and reduce their impacts. I deal with the omitted variables problem by adopting an instrumental variable approach where the instrument consists of the interaction between the propensity of a province to suffer a climate disaster (i.e., the average number of disasters in a province relative to the number of disasters in the country during the period of study) and the contemporaneous disaster variable. This estimation strategy resembles a difference-in-difference approach (DD) that compares the effect of climate disasters across provinces with different probability of suffering disasters and across times when provinces experience different number of disasters or different magnitudes of their impacts. In the 2SLS framework, the first-stage regression compares the disaster variable in more disaster-prone provinces to less disaster-prone provinces, in years with more disasters or greater impacts relative to years with less disasters or lower impacts. The second-stage regression makes a similar comparison as in the first stage with religiosity as the dependent variable. Within this DD framework, the treatment variable (i.e., the disaster variable) corresponds to a continuous treatment, rather than a binary one. I use different measures and specifications of the disaster variable to check the robustness of the results. I not only test the local effect of disasters (i.e., from disasters occurring in the province where the respondent lives), but also the country-wide effect (i.e., from disasters occurring in provinces other than the province where the respondent lives).

The results show that climate disasters have an effect on religious preferences. However, this effect affects religious preferences in different directions. Climate disasters, mainly the annual number of events and their economic costs, erode religious preferences measured by whether the person believes in God. However, human losses increase the intensity of religious preferences among religious people, measured by the frequency of attendance to religious services. These results suggest that analyzing the effect of disasters on people’s preferences requires making the distinction not only between the number and the intensity of disasters, but also between human losses and economic or material costs of disasters. The effects of climate disasters on religious preferences remain persistent across different specifications of the econometric model. I find that the change in religious preferences not only comes from disasters happening locally, but also from those happening in other provinces. However, country-wide disasters affect only the probability of believing in God, but have no effect on the intensity of religious preferences. These results suggest that climate disasters can erode religious preferences of some people, while accentuating them among those that remain religious. This paper provides evidence that despite living in an era of rapid knowledge advancement and technological change, religion provides people in Canada with a mechanism to deal with adverse experiences resulting from climate disasters.

The literature in the economics of disasters shows that these events bring economic and social disruption in society. Although some mixed evidence of the relationship between natural disasters and economic growth exists (Cavallo et al. 2013; Strobl 2012; Loayza et al. 2012; Skidmore and Toya 2002), the effect of disasters on other economic and social variables has been identified. Disasters can increase inequality (Hsiang et al. 2017), household indebtedness (Keerthiratne and Tol 2017), price levels and inflation (Parker 2017; Cavallo et al. 2014), affect education levels (Rush 2018) and human capital (Zivin et al. 2017), human health (Yonson 2018; Parida et al. 2018), agricultural output and food production (Haile et al. 2017; Vu et al. 2017), and promote conflict (Hsiang et al. 2013; Burke et al. 2009). Moreover, past experiences, the type and magnitude of disasters determine adaptation and preparedness towards future disasters (Onuma et al. 2017a, 2017b), as well as impacts on relocation (Munro and Managi 2017), migration (Eyer et al. 2018), and population displacement (Hsiang and Sobel 2016). A country’s level of income and quality of institutions (Kahn 2005), and existing natural capital (Rajapaksa et al. 2017) constitute insurance mechanisms against natural disasters.

Other research has explored the relationship between disasters and religiosity, and the role of religiosity as an informal insurance mechanism. There is evidence suggesting that natural disasters, specially earthquakes, increase religiosity (Bentzen 2015). Religious organizations can also act as ex-post insurers to help their members in times of income and consumption shocks (Chen 2010; Dehejia et al. 2007) and that church membership is related to the demand for social insurance (Ager and Ciccone 2015). In places where people expect economic risks to happen more often, religious groups can become an ex-ante insurer (Ager and Ciccone 2015). Although the psychological benefits of religion are acknowledged (Pecha and Ruprah 2015; Dehejia et al. 2007; Clark and Lelkes 2004; Gill and Lundsgaarde 2004; Smith et al. 2003; Park et al. 1990), existing papers focus mainly on religiosity as an insurance mechanism to cover material losses.

In addition to the role of religion as insurance, alternatives for mitigation and adaptation to climate change may depend to large extent on religious preferences. For instance, being religious is associated with factors that hinder mitigation, such as risk aversion (Bartke and Schwarze 2008), judging innovative ideas upon moral and ethical issues rather than scientific evidence (Gaskell et al. 2005), and supporting less enthusiastically funding for new technologies and innovation (Benabou et al. 2015; Brossard et al. 2009). Religion is nevertheless helpful to cope with adverse events and adapt to new circumstances. The importance of religious preferences for mitigation and adaptation to climate change also challenges the role of religion groups with respect to climate change. The Catholic Encyclical ‘Laudato Si’ by Pope Francis and the Islamic Declaration on Climate Change2 are examples of religious leaders calling their followers to act to reduce the causes and consequences of climate change.

My paper contributes to the literature in the economics of climate change and the economics of religion. In the first field my study provides evidence that intensifying religious preferences constitutes an adaptation mechanism to cope with human losses from disasters. The paper contributes to the second field by identifying disasters as forces that can explain both the erosion of religious preferences when the frequency and the consequences or disasters are considered, and the intensification of these preferences among religious people when disasters bring mortality and morbidity to the population. These results can have important implications for the functioning of society and, ultimately, its economic performance. For instance, researchers have shown that religious people are more trustworthy and trusting of other people, public institutions, and market outcomes (Guiso et al. 2003). They also behave less opportunistically and more cooperatively (Benabou and Tirole 2006, 2011; Levy and Razin 2012), are less willing to break the law, accept a bribe and cheat on taxes (Guiso et al. 2003). Religion also provides people with a sense of belonging to society that improves economic outcomes (Barro and McCleary 2005; McCleary and Barro 2003). Some historical events also suggest the indirect effect of natural disasters on social order and political regimes through religion (Belloc et al. 2016; Chaney 2013). My paper provides novel evidence on the different effect that material and human losses can have on religious preferences.

The rest of the paper is organized as follows. Section 2 offers background information on climate-related disasters in Canada. Section 3 presents descriptive statistics, while section 4 discusses the estimation strategy. Section 5 shows the results and Section 6 concludes.

Religiosity and Natural Disasters in Canada

Religiosity

Canada is a country traditionally seen as secular (Thiessen and Dawson 2008) where the national level of religiosity is similar to the levels observed in Western Europe (Wilkins-Laflamme 2014; Marger 2013), and much lower than the level in the US (Reimer 1995). The process of secularization of the Canadian society, measured by the number of affiliates to religious groups, has been intensifying in the last decades (Wilkins-Laflamme 2014), although the rate of decline of religiosity varies across provinces (Eagle 2011). Quebec, Alberta and British Columbia show stronger declines in affiliation levels (Wilkins-Laflamme 2014; Eagle 2011), similar to the secularization trends in Western Europe. However, it has been suggested that religiosity in Canada is driven by conviction rather than by social convention and culture, as in the case of the US (Reimer 1995). Measured by weekly and monthly attendance to religious services, religiosity has declined by 20 percentage points between 1986 and 2008 (Eagle 2011). At the same time, the number of individuals who define themselves as non-religious more than doubled in the same period, from 10.5% to 24% (Wilkins-Laflamme, 2014).

Despite the increased secularization of the Canadian society, researchers have explored the effects of religiosity on different social and economic outcomes. In terms of wellbeing, the level of religiosity is associated to positive psychological factors that contribute to mental health (Dilmaghani 2018), lower risk of depression (Balbuena et al. 2013; Mela et al. 2008), higher school connectedness among high-schoolers (Azagba et al. 2014), increased social trust (Dilmaghani 2017a, b, c) and philanthropy (Berger 2006), social integration among immigrants (Reitz et al. 2009), and higher individual wellbeing (Dilmaghani 2017a, b, c; Mela et al. 2008). Religiosity also affects economic outcomes such as level of earnings (Dilmaghani 2017a, b, c), the gender wage gap (Dilmaghani 2015), and female labor participation (Dilmaghani and Dean 2016).

Researchers have suggested two possible theories to understand the decline in religiosity in Canada. According to the secularization theory this decline can be the result of people in fact becoming non-religious, whereas the individualization theory suggests that although participation in institutionalized religious expressions is on decline, people truly remain spiritual and religious (Wilkins-Laflamme 2015; Marger 2013; Eagle 2011). Some also argue that secularization occurring in stages is the result of an improvement in living conditions in society (Wilkins-Laflamme 2014). Some individual characteristics explaining the decline in religiosity in Canada include demographic characteristics, such as race, gender and age (Balbuena et al. 2013; Marger 2013), as well as complex processes such as global migration (Connor 2008, 2009). However, secular and religious segments of the population cannot be viewed as homogeneous groups but as groups where its members show different levels of commitment to their religious beliefs and practices (Wilkins-Laflamme 2014). The identification of factors that explain the decline in religiosity in Canada is a matter of debate and remains an open question.

Natural Disasters

Canada is a relatively safe country in the world when it comes to natural disasters. The 2016 World Risk Index, classified Canada as a country with very low risk.3 However, Canadian provinces exhibit noticeable differences in the number of climate disasters, their human losses and economic costs. Information about the frequency and consequences of disasters comes from Public Safety Canada, the governmental office in charge of disaster prevention and mitigation in Canada. Public Safety Canada collects information about natural or human events that fall into the category of disaster. An event is considered as a disaster if one or more of the following criteria are met: 1) 10 or more fatalities, 2) 100 or more people affected/injured/infected/evacuated or homeless, 3) an appeal for national or international assistance, 4) historical significance, and 5) significant damage/interruption of normal processes such that the community affected cannot recover on its own. For this paper, I only consider meteorological – hydrological events (i.e., avalanches, cold events, droughts, floods, heat events, hurricanes and typhoon, storms, tornadoes, wildfires and winter storms) that are disasters related to climate conditions and affect Canada the most. Fig. 1 shows the map of Canada and the frequency of disasters in each province and territory of the country during the period 1980–2012. Provinces with a higher average number of events are Ontario with an annual average of 2.3 disasters, and Alberta with 2.2, whereas Nunavut reports the lowest annual average with 0.4 disasters. The other provinces and territories have annual averages between 0.9 and 1.9 events.
Fig. 1

Average frequency of events in Canadian provinces and territories (1980–2012)

Frequency and exposure to natural events are primary determinants of the effects of disasters in society. The other element is social vulnerability that refers to the level of preparedness of society to deal with the consequences of disasters. Both human and economic losses reflect the magnitude of the disaster and the level of vulnerability of affected locations. Fig. 2 shows the annual average number of fatalities and the annual average economic costs across provinces and territories in Canada. The information reveals that the frequency of events, human losses and the economic costs of disasters do not necessarily correlate. Provinces with more annual fatalities, on average, are Nova Scotia with 2.8 deaths, and Newfoundland and Labrador with 2.7. Ontario and Alberta, the provinces that on average experience climate disasters more often, report lower numbers of fatalities (1.5 in Alberta and 0.8 in Ontario). Manitoba and Saskatchewan are the provinces with the lowest number of fatalities per year, 0.4 and 0.2 respectively. With respect to economic costs of disasters, New Brunswick, with an annual average of 106 million of dollars, and Quebec, with 69 million of dollars, are the provinces with more economic losses, whereas the territories (Nunavut, Yukon and Northern Territories) have suffered the lowest material losses. The comparison of frequency of climate disasters and their impacts across provinces and territories of Canada shows the divergence between frequency of events, their intensity and social vulnerability measured by human and material losses.
Fig. 2

Average annual of fatalities and economic losses in Canadian provinces and territories (1980–2012)

Dataset

To study the effect of climate disasters on people’s religious preferences I use data from the International Social Survey Programme (ISSP) and from Public Safety Canada. The ISSP is an annual cross-country survey that collects information for research in social sciences. For this paper, I use information from Canadian surveys for the period 1992–2012,4 that include socio-economic and demographic characteristics of respondents and their religious preferences. I focus on the following two questions to obtain religious preferences and their intensity:
  • Religious preferences: Do you believe in God? Potential answers: Yes; no; do not respond.

  • Intensity of religious preferences: If you are a member of a religious group or church, how often do you attend religious services?

Possible answers: several times a week; once a week; 2–3 times a month; about once a month; several times a year; about once or twice a year; less than once a year; never.

The data includes 12,333 respondents, 48% are women and 81.6% define themselves as religious. The average age of respondents is 46 years, the average household size is 2.7 individuals, and the average annual family income during the period is $37,000. Table 1 shows the descriptive statistics of the variables included in the analysis. Regarding climate disasters at the province level, the annual average number of events in Canada is 1.95, annual average number of fatalities is 1.23, injured 13.52 and evacuees 1285 individuals. The annual economic costs of disaster, on average, is almost 86 million dollars (in 2012 dollars). When the annual average number of climate events is compared to the number of other natural disasters and the number of human-caused disasters (bottom lines in Table 1), it becomes clear that disasters related to climatic patterns are the most prevalent in Canada.
Table 1

Descriptive statistics of the sample and disaster variables

Variable

N

Mean

Std. Dev.

Minimum

Maximum

Annual family income ($)

12,333

36,871

30,950

0

260,000

Age of respondent (years)

12,333

46.28

16.72

16

89

Household size (members)

12,333

2.76

1.35

1

12

Annual number of events - climate

 

1.95

1.54

0

7

Annual number of fatalities - climate

 

1.23

4.24

0

39

Annual number of injured - climate

 

13.52

64.24

0

600

Economic costs (2012 millions)

 

85.90

330

0

2420

Annual number of events - natural

 

0.02

0.05

0

2

Annual number of events - human

 

0.19

0.52

0

4

Variable

N

    

Gender of respondent

12,333

Male

Female

6404

52.35%

5829

47.65%

Religion

12,333

Religious

Non-religious

9978

81.57%

2255

18.43%

Identification Strategy

Although the occurrence and intensity of climate disasters are exogenous variables for estimation purposes, human and economic losses are not. The number of fatalities or the loss of public or private assets may depend on individual or public decisions aimed at protecting the population (i.e., insurance contracts or construction regulations). Consequently, the level of preparedness of individuals and groups to deal with the adversity of disasters can be higher in disaster-prone locations. Moreover, individual preferences over risk and uncertainty may determine the location where people decide to live. The resulting correlation between preparedness or location choices and unobserved characteristics causes an endogeneity problem in the econometric estimation and biases OLS estimates. To deal with this problem, I adopt an instrumental variable (IV) approach where the instrument is constructed as the interaction between the propensity of a province to suffer a climate-related disaster (i.e., the average number of disasters relative to the national average during the period of study) and the contemporaneous disaster variable. The validity of this instrument requires the following conditions: 1) The effect of the instrument on religious preferences takes place only through its effect on the contemporaneous disaster variable (i.e., exclusion restriction). 2) The instrument has a causal effect on the contemporaneous variable. Finally, 3) the independence assumption (i.e., the instrument being as good as random) that requires that the instrument is not correlated with the error term in the estimation model.

This estimation strategy follows Nunn and Qian (2014) that adopt a difference-in-difference approach (DD) with a continuous treatment variable. In this estimation model, the first-stage regression compares the disaster variable across provinces (first difference) with different propensity to suffer disasters (more disaster-prone provinces to less disaster-prone provinces) and within provinces (second difference) in years with more or higher impacts and in years with less disasters or lower impacts. The second-stage regression is similar to the first stage and compares religiosity across and within provinces as before. Within this DD approach, the treatment variable (i.e., the disaster variable) constitutes a continuous, rather than binary, treatment. I use different measures and specifications of the disaster variable to check the robustness of the results. I not only test the local (i.e., disasters occurring in the province where the respondent lives) and the contemporaneous (i.e., disasters occurring the year before the survey) effect of disasters, but also the country-wide (i.e., disasters occurring in provinces other than the province where the respondent lives).

The estimation model, a two-stage estimation including both the reduced form (eq. 1) form and the structural equation (Eq. 2), is as follows:
$$ {Y}_{ijt}={\beta X}_{ijt}+\gamma {D}_{jt}^{\prime }+{\tau}_t+{\rho}_j+{u}_{ijt} $$
(1)
$$ {D}_{jt}^{\prime }=\alpha \left({D}_{jt-1}\ast {P}_{jt}\right)+{\beta X}_{ijt}+{\tau}_t+{\rho}_j+{\varepsilon}_{ijt} $$
(2)
Where:
Y ijt

religious outcome (religiosity and frequency of attendance)

\( {D}_{jt}^{\prime } \)

disaster variable (number, fatalities, injured, evacuees, economic losses, insurance payments)

X ijt

socio-economic characteristics (income, education, gender, age, marital status, household size, employment status, size of town)

D jt − 1

lagged disaster variable

P jt

provincial propensity to disaster (average occurrence of disasters during the period 1980–2012)

τ t

year dummies

ρ j

provincial dummies

The first dependent variable regarding religious preferences (believing in God) is binary, equal to 1 if the respondent is non-believer and 0 otherwise. The second dependent variable, the frequency of religious service attendance, is categorical, falling in one of the eight categories where 1 is going to the service very often and 8 is never attending. I transform this categorical variable following Maddala’s (1983) procedure into a continuous variable to adopt the IV approach and simplify the interpretation of the results.5

Results

In this section I present the results of different model specifications that differ in two main aspects. The first aspect corresponds to the notion of religiosity that can refer either to whether the person believes in God or to the frequency of attendance to religious services for religious people. The model specifications also differ in the aspect of the disasters that the explanatory variable of interest reflects. I include in separate regressions the annual number of disasters (i.e., frequency of events), annual number of fatalities and injured people (i.e., human losses), and the economic losses (i.e., economic costs). At the end of the section I present the results of the regression model with all disaster variables put together as explanatory variables. Other explanatory variables considered in the estimation are the linear and quadratic terms of family income, gender of respondent, education level, marital status, whether the respondent is employed, the number of members of the family, the size of the town/city where the respondent lives, and province and year dummies. These explanatory variables are included in all specifications of the estimation model that will be discussed next.6

Table 2 – Panel A presents the results for the effect of climate disasters on whether the person believes in God. Column 1 includes the frequency of events as the disaster variable, Columns 2 and 3 consider human losses, whereas Column 4 includes the economic costs of disasters. Recall that the dependent variable is binary adopting the value of 1 if the person does not believe in God, and 0 otherwise, so that explanatory variables with positive coefficients reduce the probability of believing in God. The results suggest that the frequency of events and the economic costs associated to them reduce the probability of being religious. In terms of magnitude of the effect, the values of the elasticities of the frequency of events and economic costs are 0.26 and 0.02, implying that an increase of 1 % in the number of events and in the dollar-amount of economic losses reduces the probability of being religious by 26% and 2%. The effect of human losses on religiosity shows that the number of injured people increases religiosity (i.e., a higher probability of believing in God), while the number of fatalities has no statistical effect. The elasticity of human losses, measured by the number of injured people, is 0.038, which implies that an increase in 1% in the number of injured increases the probability of being religious by almost 4%. The results also suggest that women are, on average, 21% more likely to believe in God than men, whereas the linear and square effects of family income are not statistically significant. The effects of gender and family income are consistent across model specifications.
Table 2

Determinants of religious preferences and preference intensity, main model

Panel A – Religious preferences

Belief in God

(1)

(2)

(3)

(4)

Number of disasters

Fatalities

Injured

Economic losses

Number of events (log)

0.255***

   

(0.025)

   

Number of fatalities (log)

 

0.022

  
 

(0.018)

  

Number of injured people (log)

  

−0.038*

 
  

(0.015)

 

Economic losses (log)

   

0.018***

   

(0.002)

Family income (log)

0.068

0.069

0.070

0.069

(0.036)

(0.036)

(0.036)

(0.036)

Family income square (log)

−0.005

−0.005

−0.005

−0.005

(0.003)

(0.003)

(0.003)

(0.003)

Female

−0.211***

−0.207***

−0.205***

−0.212***

(0.030)

(0.030)

(0.030)

(0.030)

Constant

−1.188***

−1.025***

−1.029***

−1.370***

(0.159)

(0.155)

(0.155)

(0.160)

Observations

12,233

12,233

12,233

12,233

Panel B – Preference intensity

Attendance

(1)

(2)

(3)

(4)

Number of disasters

Fatalities

Injured

Economic losses

Number of events (log)

−0.013

   

(0.014)

   

Number of fatalities (log)

 

−0.017

  
 

(0.009)

  

Number of injured people (log)

  

−0.024*

 
  

(0.008)

 

Economic losses (log)

   

−0.001

   

(0.001)

Family income (log)

0.005

0.005

0.004

0.005

(0.019)

(0.019)

(0.019)

(0.019)

Family income square (log)

0.000

0.001

0.001

0.001

(0.001)

(0.001)

(0.001)

(0.001)

Female

−0.107***

−0.106***

−0.107***

−0.107***

(0.016)

(0.016)

(0.016)

(0.016)

Constant

−0.689***

−0.677***

−0.672***

−0.681***

(0.090)

(0.089)

(0.089)

(0.089)

Observations

11,980

11,980

11,980

11,980

Additional controls: marital status, education level, working status, community size, province and year dummies

Standard errors in parentheses

*p < 0.05, ** p < 0.01, *** p < 0.001

Table 2 – Panel B shows the results of the regression model with the frequency of attendance to religious services as the dependent variable. The columns in this Table correspond to model specifications in Table 2 – Panel A. The question to be answered is whether climate disasters increase the frequency of attendance to religious services among respondents who believe in God. The results show that human losses increase the frequency of service attendance, while the effect of number of events and economic costs of disasters is not statistically significant. The values of the elasticity of the number of fatalities and the number of injured people are 0.017 and 0.024. The elasticity values suggest that an increase of 1% in the number of fatalities and of injured people increases the attendance to religious services in 2% and 2.5%, respectively. While family income plays no role in explaining the frequency of attendance, women are on average 11% more likely to attend religious services than men. The results in Table 2 imply that the effect of climate disasters on religiosity is twofold: the first effect is that frequency of disasters and economic losses, on average, weaken religious preferences by lowering the probability of believing in God. The second effect implies that, among religious people, human losses intensify religious preferences measured by the frequency of attendance to religious preferences. In other words, the frequency and economic costs of disasters erode religious preferences, while human losses intensify them among those professing a religion.

Payments by insurance companies can alleviate individuals’ material losses in the aftermath of disasters. Although I have no information about insurance payments at the individual level, Public Safety Canada collects information on total insurance payments per disaster. To test the effect of climate disasters on religiosity while controlling for the effect of insurance payments, I include the percentage of economic costs of disasters covered by insurance as an additional explanatory variable in the regression model. This variable is constructed as the ratio between insurance payments and economic losses so that a higher ratio implies that a higher percentage of the costs of disasters were covered by the insurance industry.

Table 3 shows the results of this model specification with religiosity and frequency of religious service attendance as dependent variables, respectively. The results in Table 3 – Panel A suggest a consistent effect of disasters on religiosity. The number of disasters and economic losses erode religiosity, whereas the number of injured people increases the probability of being religious. An increase of 1% in the number of disasters and in economic losses reduces, on average, the probability of being religious by 23.5% and 1.2%, whereas an increase of 1% in the number of injured people increases this probability by 4.3%. Regarding the ratio between insurance payments and economic costs, an increase of 1% in this ratio increases, on average, the probability of being religious by 1% to 2%.
Table 3

Determinants of religious preferences and preference intensity, including insurance payments

Panel A – Religious preferences

 

(1)

(2)

(3)

(4)

Belief in God

Number of disasters

Fatalities

Injured

Economic losses

Number of events (log)

0.235***

   
 

(0.025)

   

Number of fatalities (log)

 

0.002

  
  

(0.019)

  

Number of injured people (log)

  

−0.043*

 
   

(0.015)

 

Economic losses (log)

   

0.012***

    

(0.002)

Insurance coverage (log)

−0.015***

−0.017***

−0.017***

−0.011***

 

(0.003)

(0.003)

(0.003)

(0.003)

Family income (log)

0.069

0.071

0.071*

0.070

 

(0.036)

(0.036)

(0.036)

(0.036)

Family income square (log)

−0.005

−0.005

−0.005

−0.005

 

(0.003)

(0.003)

(0.003)

(0.003)

Female

−0.213***

−0.208***

−0.207***

−0.212***

 

(0.030)

(0.030)

(0.030)

(0.030)

Constant

−1.251***

−1.118***

−1.118***

−1.282***

 

(0.159)

(0.155)

(0.155)

(0.160)

Observations

12,233

12,233

12,233

12,233

Panel B – Preference intensity

Attendance

(1)

(2)

(3)

(4)

Number of disasters

Fatalities

Injured

Economic losses

Number of events (log)

−0.017

   
 

(0.014)

   

Number of fatalities (log)

 

−0.021*

  
  

(0.009)

  

Number of injured people (log)

  

−0.025*

 
   

(0.008)

 

Economic losses (log)

   

−0.003*

    

(0.001)

Insurance coverage (log)

−0.003*

−0.003*

−0.003*

−0.005**

 

(0.001)

(0.001)

(0.001)

(0.002)

Family income (log)

0.005

0.005

0.004

0.005

 

(0.019)

(0.019)

(0.019)

(0.019)

Family income square (log)

0.000

0.001

0.001

0.000

 

(0.001)

(0.001)

(0.001)

(0.001)

Female

−0.107***

−0.107***

−0.107***

−0.107***

 

(0.016)

(0.016)

(0.016)

(0.016)

Constant

−0.685***

−0.669***

−0.665***

−0.673***

 

(0.090)

(0.089)

(0.089)

(0.089)

Observations

11,980

11,980

11,980

11,980

Additional controls: marital status, education level, working status, community size, province and year dummies

Standard errors in parentheses

*p < 0.05, ** p < 0.01, *** p < 0.001

The results in Table 3 – Panel B show that human and economic losses increase the frequency of attendance to religious services among religious people once the insurance variable is included as an additional explanatory variable in the estimation model. The number of disasters, however, has no effect on the religious variable. The value of elasticities of the number of fatalities and injured people is 0.021 and 0.025, respectively, higher than the elasticity of economic losses, which is 0.003. The magnitude of the elasticities shows that the effect of marginal increases in human losses on the intensity of religious preferences is larger than the effect of marginal increases in the economic costs of disasters. The ratio between insurance payments and economic costs also increases the frequency of attendance, although the magnitude of the effect is relatively small (an elasticity value of 0.003 on average). These findings suggest that insurance can reduce people’s concerns about material losses and free their minds to focus more on spiritual needs. Existing evidence suggests that scarcity (e.g., scarcity resulting from disasters) can tax people’s mind and get them fixated in solving their scarcity problems while abandoning other priorities (Mullainathan and Shafir 2013). In this context, insurance seems to prevent people from getting trapped in the psychology of scarcity (conversely, consider the case of a person that suffers a big material loss and spends most of her time trying to recover what she lost, which leaves her with less time and ‘mind’ to turn to religion). The results in Table 3 also confirm that the effect of climate disasters on religiosity is persistent across model specifications after controlling for the effect of insurance payments relative to the economic costs of disasters.

The findings above suggest that disasters occurring near the place of people’s residence (i.e., in the province where the respondent lives) influence religious preferences. However, disasters happening in other provinces may also affect religiosity. To test this, I include as explanatory variables both the disaster variable for the province where the respondent lives and the disaster variable for the rest of the country. Table 4 shows the results of this model specification. Table 4 – Panel A shows that the frequency of events, human losses (concretely the number of fatalities) and the economic costs of disasters occurring in the province of residence reduce the probability of being religious. Disasters in the rest of the country (frequency, human and economic losses) also have a strong and persistent effect that decreases the probability of being religious. Disasters occurring both locally and nationally have an effect in the same direction, although the magnitude of the effect of disasters in the rest of the country, measured by elasticities, is larger than those happening locally. The value of elasticities of number of fatalities and of economic losses occurring countrywide is four and eight times the values of these elasticities for disasters occurring in the respondent’s province of residence.
Table 4

Determinants of religious preferences and preference intensity, including disasters in other provinces

Panel A – Religious preferences

Belief in God

(1)

(2)

(3)

(4)

Number of disasters

Fatalities

Injured

Economic losses

Number of events (log)

0.172***

   

(0.027)

   

Number of fatalities (log)

 

0.055**

  
 

(0.020)

  

Number of injured people (log)

  

−0.028

 
  

(0.017)

 

Economic losses (log)

   

0.009***

   

(0.003)

Disasters other provinces (log)

0.620***

0.200***

0.105***

0.075***

(0.022)

(0.018)

(0.023)

(0.004)

Family income (log)

0.065

0.074*

0.071*

0.070

(0.037)

(0.037)

(0.036)

(0.037)

Family income square (log)

−0.005

−0.005

−0.005

−0.005

(0.003)

(0.003)

(0.003)

(0.003)

Female

−0.222***

−0.207***

−0.205***

−0.220***

(0.031)

(0.030)

(0.030)

(0.031)

Constant

−2.637***

−1.234***

−1.289***

−2.626***

(0.163)

(0.155)

(0.165)

(0.171)

Observations

12,233

12,233

12,233

12,233

Panel B – Preference intensity

Attendance

(1)

(2)

(3)

(4)

Number of disasters

Fatalities

Injured

Economic losses

Number of events (log)

−0.019

   

(0.014)

   

Number of fatalities (log)

 

−0.017

  
 

(0.009)

  

Number of injured people (log)

  

−0.023**

 
  

(0.008)

 

Economic losses (log)

   

−0.002

   

(0.001)

Disasters other provinces (log)

0.015

0.006

0.008

0.003

(0.010)

(0.008)

(0.009)

(0.002)

Family income (log)

0.004

0.005

0.004

0.004

(0.019)

(0.019)

(0.019)

(0.019)

Family income square (log)

0.001

0.001

0.001

0.001

(0.001)

(0.001)

(0.001)

(0.001)

Female

−0.107***

−0.107***

−0.107***

−0.000

(0.016)

(0.016)

(0.016)

(0.000)

Constant

−0.662***

−0.665***

−0.672***

−0.642***

(0.092)

(0.091)

(0.089)

(0.092)

Observations

11,980

11,980

11,980

11,980

Additional controls: marital status, education level, working status, community size, province and year dummies

Standard errors in parentheses

*p < 0.05, ** p < 0.01, *** p < 0.001

Regarding the effect of disasters on the intensity of religious preferences, Table 4 – Panel B shows that human losses, specifically the number of injured people, increase the frequency of attendance to religious services among religious people. The effects of the number of events and economic costs of disasters, as well as the effects of disasters occurring in the rest of the country are not statistically significant. These results suggest that the disasters that affect the intensity of religious preferences are those happening locally in the province of residence of respondents, whereas those occurring nationally have a much less important effect. Possible explanations for this are the following: 1) religious people get more motivated to attend special religious services convocated as a result of disasters happening nearby, 2) these special religious services are more often convocated in places where disasters occur, or 3) these two possibilities happening at the same time.

The results discussed in this section suggest that the effect of climate disasters on religious preferences is not unidirectional and depends on how religiosity is defined and what aspect of disasters is considered. When the number of events and the economic costs of disasters are considered, climate disasters reduce the probability for the average person to be religious, which constitutes an erosion of religious preferences. In contrast, human losses of climate disasters intensify religious preferences among religious people, measured as the frequency of attendance to religious services. As discussed above, these results are consistent across different specifications of the estimation model. The channels through which climate disasters affect religious preferences remain open for future research work. Whether people becoming less religious as disasters increase in number and material losses is the result of having more scientific information that links climate disasters to human-induced climate change, or whether climate disasters reaffirm religious people’s beliefs that disasters are acts of God and that He will protect them, are some potential explanations. Moreover, whether changes in religiosity that result from climate disasters promote or prevent people from taking measures to mitigate and adapt to climate change requires further work.

Table 5 summarizes the results from the first stage of the instrumental variable estimations for the basic model, the model including insurance payments, and the model considering disasters in other provinces. The estimates show that the effect of the instrument on the instrumented variable is statistically significant. A complete presentation of the results from the first-stage estimation is shown in Appendix Tables 9 and 10.
Table 5

Summary of first-stage estimates for instrumental variable regressions

 

Dependent

variable

Disaster variable

Instrumental variable

 

Number

Fatalities

Injured

Costs

Model 1

Basic model

Belief in God

1.247 ***

1.320 ***

1.122 ***

1.284 ***

(0.003)

(0.006)

(0.004)

(0.003)

Attendance

1.250 ***

1.307 ***

1.119 ***

1.288 ***

(0.003)

(0.005)

(0.004)

(0.003)

Model 2

With insurance payments

Belief in God

1.234 ***

1.314 ***

1.122 ***

1.221 ***

(0.003)

(0.006)

(0.004)

(0.003)

Attendance

1.237 ***

1.301 ***

1.119 ***

1.222 ***

(0.003)

(0.005)

(0.004)

(0.003)

Model 3

With disasters in other provinces

Belief in God

1.205 ***

1.322 ***

1.120 ***

1.230 ***

(0.003)

(0.006)

(0.004)

(0.003)

Attendance

1.215 ***

1.310 ***

1.112 ***

1.241 ***

(0.003)

(0.005)

(0.004)

(0.003)

Standard errors in parentheses

*p < 0.05, ** p < 0.01, *** p < 0.001

Finally, when the disaster variables are put together in the right-hand side of the regression model, the differentiated effect of human and economic losses is confirmed. Table 6 shows the results of this model where believing in God and attendance to religious service are the dependent variables. In column 1, the number of events and the economic costs reduce the probability of being religious, while the number of injured people increases this probability. As before, the economic costs of disasters erode religiosity, whereas human losses make people belief in God more. Regarding the intensity of the religious preferences, only human losses (specifically the number of injured and evacuees) have a statistically significant effect that increases the frequency of attendance to religious services.
Table 6

Regression model with all disaster variables as independent variables

 

(1)

(2)

Belief in God

Attendance

Number of events (log)

0.217***

0.004

 

(0.030)

(0.017)

Number of fatalities (log)

0.011

0.017

 

(0.025)

(0.012)

Number of injured people (log)

−0.095***

−0.027**

 

(0.019)

(0.009)

Number of evacuees (log)

−0.000

−0.012***

 

(0.006)

(0.003)

Economic cost (log)

0.014***

0.002

 

(0.003)

(0.002)

Family income (log)

0.068

0.005

 

(0.036)

(0.019)

Family income square (log)

−0.005

0.000

 

(0.003)

(0.001)

Female

−0.214***

−0.107***

 

(0.030)

(0.016)

Household size

0.002

0.002*

 

(0.001)

(0.001)

Constant

−1.354***

−0.631***

 

(0.161)

(0.144)

Observations

12,233

11,992

Additional controls: marital status, education level, working status, community size, province and year dummies

Standard errors in parentheses

*p < 0.05, ** p < 0.01, *** p < 0.001

Concluding Remarks

Canada is a relatively disaster-safe country, with adequate infrastructure to reduce the economic costs of disasters and well-functioning institutions that allows people to adapt or minimize material losses from catastrophic events (e.g., insurance markets and emergency funds). However, the findings of the paper suggest that the losses imposed by disasters motivate people to look for alternatives to ease the potential pain, anxiety or stress that result from catastrophic losses. Either directly or indirectly affected by a disaster, people feel the need to turn to mechanisms to cope specially with human losses. This paper offers evidence that religion is one of these mechanisms.

Although the effect of climate disasters can be expected to have a homogenous effect on religious preferences, this paper reveals that the frequency of disasters and the related human and economic losses have different effects. Concretely, the frequency of disasters and their economic costs erode religious preferences measured by the probability that a person believes in God. Human losses, defined in terms of the number of fatalities and of injured people, increase the intensity of religious preferences among religious people, measured by the frequency of religious service attendance. These results are robust to different specifications of the regression model. Moreover, climate disasters have an effect on religious preferences not only when events happen locally (i.e., in the province where respondents live), but also when disasters happen in other provinces. More work is needed to understand the temporal dimensions of the effect of disasters, that refer to whether the contemporaneous (i.e., the effect of disasters occurring the previous year) or the cumulative effects of disasters (i.e., the effect of disasters occurring during the four or five previous years) have stronger impacts on religiosity.

Since the predominant religion among Canadian people is Christianity (about 90% of people belong to Catholicism, Protestantism, or other Christian groups), the results of the paper cannot be extended to other countries with different predominant religions or people professing other religious faiths. Although this constitutes a limitation of the analysis, the results confirm the role of religion as a means for people to cope with adverse situations. Since Canada has institutions and mechanisms to deal with the consequences of natural disasters, the findings of the paper may also suggest that the effect of natural disasters on religiosity in the rest of the world, and especially in the developing world where such institutions and mechanisms are absent, might be stronger. In the context of countries with weak institutions and malfunctioning financial and insurance markets, religion may become one of the most important sources of relief for individuals going through adversity in the context of climate change.

Changes in religiosity that result from climate disasters can have important implications for climate change policy. Whether religiosity and mitigation or adaptation to climate change are complementary or substitutes goods can determine the efficacy of this type of policy. If religiosity and private insurance are substitutes, economic losses from climate disasters could be higher in the future. But if religiosity and social capital are complements, more social adaptation and cooperation can be expected as a result of this type of disasters. Or if religiosity and individual preferences over public expenditure are interrelated, changes in religiosity can influence the implementation of policies aimed at improving distribution outcomes in society. These issues are just a few examples of topics that arise from the study of the effect of disasters on individual preferences, that remain as open questions and that require further research work.

Footnotes

  1. 1.

    In Canada, the cases of the Calgary flood and Ontario floods in 2013 and 2014, and the case of wildfires in Fort McMurray. In the US, the cases of Hurricanes Katrina and Sandy in 2005, and 2012.

  2. 2.

    Both released in summer 2015.

  3. 3.

    The World Risk Index (www.irdrinternational.orf/2016/03/01/word-risk-index/) measures the vulnerability of society to natural disasters. The index defines 5 risk categories: very low (from a value of 0.08 to 3.46), low (from 3.47 to 5.46), middle (from 5.47 to 7.30), high (from 7.31 to 10.39), and very high (from 10.40 to 36.72). The value of the Index for Canada is 3.01% (very low risk). According to this Index, the safest country is Qatar with a value 0.08%, while the riskiest is Vanuatu with a value of 36.72% (very high risk). For comparison purposes, the US has a value in the ranking of 3,76% (low risk), Great Britain a value of 3.54% (low risk), and Australia a value of 4.22% (low risk). The riskiest areas of the world include Southeast Asia, Central America and the Southern Sahel.

  4. 4.

    With the exception of four years (2002, 2007, 2008 and 2009) when Canada was not part of the survey.

  5. 5.

    The results from Ordered Probit and Interval Regressions produce similar results.

  6. 6.

    The results from the estimation models that disregard the problem of endogeneity are presented in Appendix Tables 7 and 8. Significant and stronger effects are found when endogeneity is considered.

References

  1. Ager P, Ciccone A (2015) Agricultural risk and the spread of religious communities”, EHES Working Papers in Economic History, No.74.Google Scholar
  2. Azagba S, Asbridge M, Langille DB (2014) Is religiosity positively associated with school connectedness: Evidence from High School Students in Atlantic Canada. J Prim Prev 35(6):417–427CrossRefGoogle Scholar
  3. Balbuena L, Baetz M, Bowen R (2013) Religious attendance, spirituality, and major depression in Canada: A 14-year follow-up study. Can J Psychiatr 58(4):225–232CrossRefGoogle Scholar
  4. Barro R, McCleary RM (2005) Which countries have state religions? Q J Econ 120(4):1331–1370CrossRefGoogle Scholar
  5. Bartke S, Schwarze R (2008) Risk-averse by nation or by religion? some insights on the determinants of individual risk attitudes. SOEP paper No. 131Google Scholar
  6. Belloc M, Drago F, Galbiati R (2016) Earthquakes, religion, and transition to self-government in Italian cities. CESifo Working Paper Series No. 5566Google Scholar
  7. Benabou R, Tirole J (2006) Incentives and prosocial behavior. Am Econ Rev 96(5):1652–1678CrossRefGoogle Scholar
  8. Benabou R, Tirole J (2011) Identity, morals, and taboos: beliefs as assets. Q J Econ 126(2):805–855CrossRefGoogle Scholar
  9. Benabou R, Ticchi D, Vindigni A (2015) Forbidden fruits: the political economy of science, religion, and Growth. NBER Working Paper No. 21105. Cambridge. MAGoogle Scholar
  10. Bentzen JS (2015) Acts of God? religiosity and natural disasters across subnational world districts. University of Copenhagen, Department of Economics Discussion Paper No. 15–06.Google Scholar
  11. Berger IE (2006) The influence of religion on philanthropy in Canada. Int J Volunt Nonprofit Org 17(2):115–132Google Scholar
  12. Brossard D, Scheufele DA, Kim E, Lewenstein BV (2009) Religiosity as a perceptual filter: examining processes of opinion formation about nanotechnology. Public Underst Sci 18(5):546–558CrossRefGoogle Scholar
  13. Burke MB, Miguel E, Satyanath S, Dykema JA, Lobell DB (2009) Warming increases the risk of civil war in Africa. PNAS 106(49):20670–20,674CrossRefGoogle Scholar
  14. Cavallo E, Galiani S, Noy I, Pantano J (2013) Catastrophic natural disasters and economic growth. Rev Econ Stat 95(5):1549–1561CrossRefGoogle Scholar
  15. Cavallo A, Cavallo E, Rigobon R (2014) Prices and supply disruptions during natural disasters. Rev Income Wealth 60(S2):S449–S471.  https://doi.org/10.1111/roiw.12141 CrossRefGoogle Scholar
  16. Chaney E (2013) Revolt on the Nile: economic shocks, religion and political power. Econometrica 81:2033–2053CrossRefGoogle Scholar
  17. Chen DL (2010) Club goods and group identity: evidence from islamic resurgence during the indonesian financial crisis. J Polit Econ 118:300–354CrossRefGoogle Scholar
  18. Clark A, Lelkes O (2004) Deliver us from evil: religion as insurance. Papers on economics of religion 06/03.Google Scholar
  19. Connor P (2008) Increase or decrease? the impact of the international migratory event on immigrant religious participation. J Sci Study Relig 47(2):243–257CrossRefGoogle Scholar
  20. Connor P (2009) Immigrant religiosity in Canada: Multiple trajectories. J Int Migr Integr 10(2):159–175CrossRefGoogle Scholar
  21. Dehejia R, DeLeire T, Luttmer EFP (2007) Insuring Consumption and Happiness through Religious Organizations. J Public Econ 91:259–279CrossRefGoogle Scholar
  22. Dennett DC (2006) Breaking the Spell: Religion as Natural Phenomenon. Penguin Group, New YorkGoogle Scholar
  23. Diamond J (2012) The world until yesterday: What can we learn from traditional societies? Penguin Group, New YorkGoogle Scholar
  24. Dilmaghani M (2015) Religiosity, gender, and wage: the differentiated impact of private prayer in Canada. Int J Soc Econ 42(10):888–905CrossRefGoogle Scholar
  25. Dilmaghani M (2017a) Religiosity and labour earnings in Canadian Provinces. J Lab Res 38(1):82–99CrossRefGoogle Scholar
  26. Dilmaghani M (2017b) Religiosity and subjective Wellbeing in Canada. J Happiness Stud.  https://doi.org/10.1007/s10902-016-9837-7
  27. Dilmaghani M (2017c) Religiosity and social trust: evidence from Canada. Rev Soc Econ 75(1):49–75CrossRefGoogle Scholar
  28. Dilmaghani M (2018) Importance of religion or spirituality and mental health in Canada. J Relig Health 58(1):120–135CrossRefGoogle Scholar
  29. Dilmaghani M, Dean J (2016) Religiosity and female labour market attainment in Canada: the Protestant exception. Int J Soc Econ 43(3):244–262CrossRefGoogle Scholar
  30. Eagle DE (2011) Changing patterns of attendance at religious services in Canada, 1986–2008. J Sci Study Relig 50(1):187–200CrossRefGoogle Scholar
  31. Eyer J, Dinterman R, Miller N, Rose A (2018) the effect of disasters on migration destinations: Evidence from Hurricane Katrina. Econ Disasters Clim Chang.  https://doi.org/10.1007/s41885-017-0020-3
  32. Gaskell G, Hallman W, Jackson J (2005) Social values and the governance of science. Science 310:1908–1909CrossRefGoogle Scholar
  33. Gill A, Lundsgaarde E (2004) State welfare spending and religiosity: A cross-national analysis. Ration Soc 16(4):399–346CrossRefGoogle Scholar
  34. Guiso L, Sapienza P, Zingales L (2003) People’s opium? religion and economic attitudes. J Monet Econ 50(1):225–282CrossRefGoogle Scholar
  35. Haile MG, Wossen T, Tesfaye K, von Braun J (2017) Impact of climate change, weather extremes, and price risk on global food supply. Econ Disasters Clim Chang 1(1):55–75.  https://doi.org/10.1007/s41885-017-0005-2 CrossRefGoogle Scholar
  36. Hoeppe P (2016) Trends in weather related disasters: Consequences for insurers and society. Weather Clim Extremes 11:70–79CrossRefGoogle Scholar
  37. Hsiang SM, Sobel AH (2016) Potentially extreme population displacement and concentration in the tropics under non-extreme warming. Sci Rep.  https://doi.org/10.1038/srep25697
  38. Hsiang SM, Burke M, Miguel E (2013) Quantifying the influence of climate on human conflict. Science 341(6151):1235367.  https://doi.org/10.1126/science.1235367 CrossRefGoogle Scholar
  39. Hsiang SM, Kopp R, Jina A, Rising J, Delgado M, Mohan S, Rasmussen DJ, Muir-Wood R, Wilson P, Oppenheimer M, Larsen K, Houser T (2017) Estimating economic damage from climate change in the United States. Science 356(6345):1362–1369.  https://doi.org/10.1126/science.aal4369 CrossRefGoogle Scholar
  40. Iannaccone LR (1988) A Formal Model of Church and Sect. Am J Sociol 94(Supplement):S241–S268CrossRefGoogle Scholar
  41. Iannaccone LR (1992) Sacrifice and Stigma: Reducing Free-Riding in Cults, Communes, and Other Collectives. J Polit Econ 100(2):271–291CrossRefGoogle Scholar
  42. Iannaccone LR (1995) Voodoo Economics? Reviewing the Rational Choice Approach to Religion. J Sci Study Relig 34(1):76–88CrossRefGoogle Scholar
  43. Kahn ME (2005) The Death Toll from Natural Disasters: The Role of Income, Geography and Institutions. Rev Econ Stat 87(2):271–284CrossRefGoogle Scholar
  44. Keerthiratne S, Tol RSJ (2017) Impact of Natural Disasters on Financial Development. Econ Disasters Clim Chang 1:33–54CrossRefGoogle Scholar
  45. Kneale M (2014) An Atheist’s History of Belief: Understanding our Most Extraordinary Invention. The Bodley Head. London.Google Scholar
  46. Levy G, Razin R (2012) Religious Beliefs, Religious Participation, and Cooperation. Am Econ J: Microecon 4(3):121–151Google Scholar
  47. Loayza NV, Olaberría E, Rigolini J, Christiaensen L (2012) Natural Disasters and Growth: Going Beyond the Average. World Dev 40(7):1317–1336CrossRefGoogle Scholar
  48. Maddala GS (1983) Limited Dependent and Qualitative Variables in Econometrics. Cambridge: Cambridge University PressGoogle Scholar
  49. Marger M (2013) Religiosity in Canada and the United States: Diverging Paths. Am Rev Can Stud 43(1):70–85CrossRefGoogle Scholar
  50. McCleary RM, Barro R (2003) Religion and economic growth across countries. Am Sociol Rev 68(5):760–781CrossRefGoogle Scholar
  51. Mela MA, Marcoux E, Baetz M, Griffin R, Angelski C, Deqiang G (2008) The effect of religiosity and spirituality on psychological well-being among forensic psychiatric patients in Canada. Ment Health Relig Cult 11(5):517–532CrossRefGoogle Scholar
  52. Mullainathan S, Shafir E (2013) Scarcity: why having too little means so much (1st ed.). Times Books, Henry Holt and Co. New York.Google Scholar
  53. Munro A, Managi S (2017) Going back: radiation and intentions to return among household evacuated after the great tohoku earthquake. Econ Disasters Clim Chang 1(1):77–93.  https://doi.org/10.1007/s41885-017-0001-6 CrossRefGoogle Scholar
  54. Nunn N, Qian N (2014) US Food Aid and Civil Conflict. Am Econ Rev 104(6):1630–1666CrossRefGoogle Scholar
  55. Onuma H, Shin KJ, Managi S (2017a) Household preparedness for natural disasters: Impact of disaster experience and implications for future disaster risks in Japan. Int J Disaster Risk Reduct 21:148–158CrossRefGoogle Scholar
  56. Onuma H, Shin KJ, Managi S (2017b) Reduction of future disaster damages by learning from disaster experiences. Nat Hazards 87(3):1435–1452CrossRefGoogle Scholar
  57. Parida Y, Dash DP, Bhardwaj P, Chowdhury JR (2018) Effects of drought and flood on farmer suicide in Indian States: an empirical analysis. Econ Disasters Climat Chang.  https://doi.org/10.1007/s41885-018-0023-8
  58. Park C, Cohen L, Herb L (1990) Intrinsic religiousness and religious coping as life stress moderators for Catholics versus Protestants. J Pers Soc Psychol 59(3):562–574CrossRefGoogle Scholar
  59. Parker M (2017) The Impact of Disasters on Inflation. Econ Disasters Clim Chang.  https://doi.org/10.1007/s41885-017-0017-y
  60. Pecha C, Ruprah IJ (2015) Religion as an Unemployment Insurance and the Basis of Support for Public Safety Nets: The Case of Latin America and the Caribbean. IDB Working Paper Series No. 601. Washington D.C.Google Scholar
  61. Rajapaksa D, Islam M, Managi S (2017) Natural capital depletion: the impact of natural disasters on inclusive growth. Econ Disasters Climat Chang.  https://doi.org/10.1007/s41885-017-0009-y
  62. Reimer SH (1995) A look at cultural effects on religiosity: A comparison between the United States and Canada. J Sci Study Relig 34(4):445–457CrossRefGoogle Scholar
  63. Reitz JG, Barnerjee R, Phan M, Thompson J (2009) Race, religion, and the social integration of new immigrant Minorities in Canada. Int Migr Rev 43(4):695–726CrossRefGoogle Scholar
  64. Rush JV (2018) The impact of natural disasters on education in indonesia. Econ Disasters Climat Chang.  https://doi.org/10.1007/s41885-017-0022-1
  65. Scheve K, Stasavage D (2006) Religion and preferences for social insurance. Q J Pol Sci 1:255–286CrossRefGoogle Scholar
  66. Skidmore M, Toya H (2002) Do natural disasters promote long-run growth? Econ Inq 40(4):664–687CrossRefGoogle Scholar
  67. Smith TT, Hardman RK, Richards PS, Fischer L (2003) Intrinsic religiousness and spiritual well-being as predictors of treatment outcome among women with eating disorders. The Journal of Treatment and Prevention 11(1):15–26Google Scholar
  68. Stern G (2007) Can God Intervene? How Religion Explains Natural Disasters. Praeger Publishers, WestportGoogle Scholar
  69. Strobl E (2012) The economic growth impact of natural disasters in developing countries: Evidence from hurricane strikes in the Central American and Caribbean regions. J Dev Econ 97(1):130–141CrossRefGoogle Scholar
  70. Thiessen J, Dawson LL (2008) Is there a “renaissance” of religion in Canada? A critical look at Bibby and beyond. Stud Religi 37(3–4):389–415CrossRefGoogle Scholar
  71. Vu TB, Im EI, Hayashi K, Torio R (2017) Cyclones, Deforestation, and Production of Food Crops in Vietnam. Econ Disasters Climat Chang 1(3):245–262.  https://doi.org/10.1007/s41885-017-0010-5 CrossRefGoogle Scholar
  72. Wilkins-Laflamme S (2014) Towards Religious Polarization? Time Effects on Religious Commitment in U.S., UK, and Canadian Regions. Sociol Relig 75(2): 284–308Google Scholar
  73. Wilkins-Laflamme S (2015) How unreligious are the religious “Nones”? Religious Dynamics of the Unaffiliated in Canada. Can J Sociol 40(4):477–500Google Scholar
  74. Wilson EO (2012) The Social Conquest of Earth. Liveright Publishing Corporation, New YorkGoogle Scholar
  75. Yonson R (2018) Floods and pestilence: diseases in Philippine urban areas. Econ Disasters Climat Chang.  https://doi.org/10.1007/s41885-017-0021-2
  76. Zivin G, Hsiang JS, Neidell M (2017) Temperature and human capital in the short and long-run. J Assoc Environ Resour Econ.  https://doi.org/10.1086/694177

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Water Planning Lab – School of Community and Regional Planning (SCARP)University of British ColumbiaVancouverCanada

Personalised recommendations