RELIGIOUSITY AND
SELF-EMPLOYMENT IN
THE UNITED KINGDOM
By: Paul Korver VU ID: 1919547 UvA ID: 10887725 e-mail: prkorver@hotmail.com Supervisor: MSc. S.F.W. (Fleur) Meddens Second reader: prof. dr. P.D. (Philipp) Koellinger
1
Abstract
This study focuses on the relationship between religion and the chance to be self-employed within the UK. In this study religion has been divided into three groups: Christianity, Islam and other religions. Because the ethnic background of a person plays a large role on the religion one belongs to, because each country has its main religion and most people belong to the religion that is the main religion in the country where their ethnicity lies, it is used as a control variable. Evidence is found for relationships between all religious variables and the chance to be self-employed. Results show that Christians have a lower chance to be self-employed, compared to non-Christians. It is suggested that this is linked to the values that belong to being Christian, and that these results might be different for very devout Christians. The relationships between self-employment and both Muslims and members of other religions (not Christians or Muslims) on the other hand show a higher chance to be self-employed, compared to non-members of the respective groups. Suggested explanation for the higher chance of Muslims to become self-employed is that they are discriminate against, and therefore become necessity entrepreneurs. However as ethnicity is being controlled for they are being discriminate against because of their religion and not because of their non-white ethnicity. Unexpectedly the relationship between people from non-white ethnicity and employment results in a lower chance to be self-employed. This suggests that ethnicity does not play a large role when it comes to the relationship between religion and the chance to be self-employed.
2
Table of Content
Abstract ... 1 Table of Content ... 2 Introduction ... 4 Literature review ... 5 Christianity ... 6 Islam ... 7 Other Religions ... 9 Ethnic Background ... 10 Method ... 12 Data ... 12 Sample ... 12 Study materials ... 13 Statistical Power ... 13 Dependent variable ... 14 Independent variables ... 14 Control variables ... 15 Correlation ... 15 Descriptives ... 15 Model ... 16 Results ... 17 Correlation ... 18 Logistic regression ... 21 Regression ... 22Assessing the model ... 23
3
Summary of results ... 26
Interpretation of the results ... 26
Limitations ... 29
Future research ... 30
4
Introduction
Since at least 1905 (Weber, 1930) religion has been linked to economic performance, and an important part of that relationship between religion and economic growth can be attributed to the relationship between religion and self-employment (Audretsch & Keilbach, 2004). Today religion still is an important aspect in many people’s lives as it can affect many choices one makes. This way religion can lead to effect behavioural aspects like: values, beliefs and economic decision making, although the causality can only be strongly suggested (Iannaccone, 1998). In a more practical sense this is seen in the choice of marital partner, education, vocation and career choice, which do show a causal relationship between religion and these practical aspects (Lehrer, 2004). In turn these behavioural and practical implications of religion will influence important economic factors such as wealth and wages. The role religion plays in shaping economic behaviour has received growing attention in the past years. Some studies have found a correlation between religion and the positive effect of some religious aspects on economic performance, like higher returns on investments in human capital or even higher income (Chiswick, 1983; Steen, 1996), and some have found religion to be negatively correlated to economic performance, like the negative influence correlated between religions and income in the case of countries where the average income is low (Eastern Europe), or even as a negative relationship between church attendance and economic growth found within various countries across all continents (including the U.K., the U.S., Australia, China, Russia, Netherlands and Argentina) (Barro & McCleary, 2003; Bettendorf & Dijkgraaf, 2010). Both the positive and the negative relationship of religion with economic performance are explained by the enduring aspects of religiosity.
An important factor that could potentially influence the relationship between religion and economic development is the country the test is conducted in, because whether a religion is a majority or minority religion within the country the test is conducted in can potentially influence the relationship found between the religion and self-employment . The type of religion (majority or minority) per country could affect the relationship between the religions and economic development as the discrimination against minority religions could potentially influence the chance of becoming self-employed as could the restrictions of majority religions (Barro & McCleary, 2003). However while controlling for the differences of countries Guiso et al. (2003) found on average a higher per capita income and growth associated with religious belief of Christians, Jews, Muslims, Buddhists and Hindus. This could potentially indicate that religion itself does have a relationship with economic performance.
5 Studying the relationship between religion and self-employment can help explain the differences in productivity and higher per capita income and growth (Caree & Thurik, 2003). Although a positive relationship between religion and self-employment has already been found in some studies, case and literature studies conducted on different religions, including Christianity, Islam and Judaism, and with information about different countries, including Asia Europe and the United States (Dodd & Gotsis, 2007; Dana, 2009), there is a lack of generalizable quantitative studies on the relationship. Most studies on the relationship between religion and self-employment are case studies conducted on selected business owners or literature studies focussed on the available, mostly case, studies. The available quantitative studies (Audretsch et al., 2013; Park et al., 2014; Zelekha et al., 2014) focus on non-generalizable samples, like the population of India or a constructed LinkedIn database. However as in the relationship between religion and self-employment within Western Europe appeals more to me, for which the previous mentioned studies are not generalizable and other quantitative studies could not be found, this study focusses on Western Europe and the UK specifically. Furthermore the limited amount of empirical research findings show a high amount of variety which could potentially be explained by factors such as type of religions researched and whether or not the religion is a majority religion in the country where the research is done, which differ from study to study. This difference in countries and religion can influence the research findings because the culture and economic status of the country, both of which could be correlated to religion and employment, also influences the choice to become employed. For religion it is expected that each religion has its own relationship with self-employment, dependent on the values and standards of the religion. However all findings share the same difficulty in measuring the causality of the effect of religion (Benjamin et al., 2010). This difficulty can be explained by the lack of information available on people who change or choose a religion later in life.
Literature review
Audretsch et al. (2013) conducted a study under almost eighty two thousand Indians on the relationship between religion and the chance to be self-employed. By the use of a questionnaire, information was collected on the characteristics of the respondents, their economic activities and their religion. Results show that Islam and Christianity have a positive relationship with the chance to be self-employment, while other religions (Buddhism, Hinduism and Sikhism) showed a negative relationship with the chance to be self-employment. However since this study was done in India, where another religion is the majority religion (Hinduism) and thus
6 other religions are minority religions than in the UK (where Christianity is the majority religion) and because the caste system still has a strong presence in India, the results are hard to generalize to other countries.
Furthermore, in another study on the relationship between religion and self-employment, Zelekha et al. (2014) found matching results for Christianity and Islam, but also found a positive relation between Buddhism and Hinduism and the chance to be self-employed people, when conducting a comparable study using a special database created with information gathered through the social network website LinkedIn. The data acquired through LinkedIn formed a database of the entrepreneurial activity of 176 countries. However considering the database is constructed with LinkedIn, which is a social network website, the information gathered is likely to be skewed towards discipline where the usage of this platform is the most beneficial and might under represent the people who do not use LinkedIn while they might benefit from it (e.g. older people or people that do not know it exists). Therefore the sample created in this manner is not likely to represent the population of the 176 countries correctly.
As mentioned earlier the differences in the relationship between religion and self-employment is expected when measuring in different countries or differently constructed databases.
Christianity
Because of the little empirical evidence on the different relationships between individual religions and self-employment, studies on the individual religions and their relationship with employment are reviewed to understand how the religions relationship with self-employment could be explained.
In the quantitative studies of Audretsch et al. (2013) and Zelekha et al. (2014) described earlier, evidence was found that Christianity increased the chance to be self-employed, compared to non-Christians. Research on the relationship between individual religions and the motivation to be self-employed reveals a positive relationship between Christianity and the motivation to be self-employed (Griebel et al., 2014; Rietveld & van Burg, 2014) when Christianity was the main religion of the country the research was conducted in. The main reason given by Griebel et al. (2014) was that they found that the twenty one Christian entrepreneurs interviewed for the study could not sufficiently combine their religious beliefs and their work identity within their previous jobs. Within their own businesses they could combine their beliefs and their work identity as they themselves saw fit. The possibility to create this workplace where they could combine these values was the main motivator to become
self-7 employed for these entrepreneurs and to quit their previous wage job. This motivation can potentially be a main driver for the higher chance to be self-employed for Christians, compared to non-Christians, reported in the quantitative studies. The sampling for the twenty-one entrepreneurs Christians, from Colorado in the United States, happened through snowball sampling and also consisted of entrepreneurs whom where not self-employed but rather displayed entrepreneurial behaviour. This resulted in respondents of various ages and various types of Christianity.
Rietveld & van Burg (2014) conducted quantitative study consisted of 756 Dutch protestant Christians who were members of Christian trade unions. In this study it was found that protestant Christian entrepreneurs have a stronger feeling their work is a calling from God, compared to protestant Christian employees, and that it is their duty to add value to the society through their work related activities. The values vocation and societal service showed positive relationships with all lines of work, however the positive relationship between entrepreneurship and Christianity has the largest impact of these relationships. These different factors all potentially indicate an increased chance to be self-employed for Christians, compared to non-Christians. In other words these two studies do not see eye to eye on the reason of self-employment of Christians, Griebel et al. (2014) indicating that freedom of religious expression was the key driver for employment while Rietveld & van Burg (2014) indicate that the self-employment comes from a so called calling from God and thus intrinsic motivation is the key driver. However both study’s findings indicate factors that potentially could positively affect the relationship between being Christian and the chance and motivation to be self-employed, compared to non-Christians. Therefore, given the current literature, it is expected that in this study Christianity will also give a positive relation with chance to be self-employed.
Islam
In the quantitative studies of Audretsch et al. (2013) and Zelekha et al. (2014) described earlier, evidence was found that being Muslim increased the chance to be self-employed, compared to non-Muslims. This relationship between being a Muslim and the motivation to be self-employed has also been found in studies focussed only on the Islam (Essers & Benschop, 2009; Pistrui & Fahed-Sreih, 2010). Esser & Benschop (2009) used a qualitative study on Muslim women who are self-employed in the Netherlands. The sample gathered through snowball sampling consisted of twenty Turkish and Moroccan women. The analysis of their life stories showed that their interpretation of the Qur’an had helped to recognize businesses opportunities in unexplored niches. This was possible because Muslims with a very traditional interpretation
8 of the Qur’an and the Islam would not have thought it to be appropriated for these women to start their own businesses. Therefore the competition in the niche markets was very little, as for example one of the women started a driving school for other Muslim women. This access to unexplored business niches can potentially increases the motivation for these Muslim women to be self-employed, compared to non-Muslim women. The women point out that they separate their religious background from their ethnic background. Because the ethnic background does not give any grounds for women to become self-employed the women distance themselves from their ethnicity in favour of their religious background. Therefore for the self-employed women in this study their religious background did not seem to be correlated with their ethnic background. The context of this study, being conducted on immigrants from Turkey and Morocco in the Netherlands, make the results of this study applicable for the study on the relationship between the Islam and self-employment in the UK.
Subsequently, factors that potentially can have a positive effect on the relationship between the Islam and the self-employment were also found for countries where the Islam is the dominant religion (the Middle East and particularly the Gulf region) (Pistrui & Fahed-Sreih, 2010). With a study of the literature landscape Pistrui & Fahed-Sreih show how the Islam provides an environment that is advantageous for becoming self-employed. The main advantage comes from the specific from of finance. In the countries where the Islam is the majority religion a form of finance called Islamic finance is emerging. Since the Islam forbids lending and borrowing money with interest there are alternative means to financing a business, mainly risk sharing. This enables a more accessible form of finance for entrepreneurs. As of 2014 the UK had six “fully Islamic” banking organisations (Breibart London, 2014). Therefore the advantage of easier access to finance could also affect the self-employment of Muslims in the UK. This form of finance works especially well within the Islam because the profit made from the interest-free investments is not only permitted by the Islam, it is also encouraged. Following from sayings attributed to the Prophet the encouragement of entrepreneurship in the Islam becomes clear: “The best gains from honourable trade and from a man’s work with his own hands”, and “To seek lawful gain is the duty of every Muslim” (Lewis and Churchill, 2009). Furthermore Muslims have a high amount of family businesses within countries where the Islam is the main religion (e.g. Turkey, North-Cyprus and Pakistan), compared to other majority religions in their respective countries (Hindus from India or Christians from Mozambique), a trait they are likely to continue to display when immigrating to other countries (Basu, 2004). Following the results of these studies it is expected that Islam has a positive relationship with the amount of self-employment within the boundaries of this study.
9
Other Religions
Because of the large amount of different religions but their limited presence in the UK and the in the sample (see next chapter) the remaining religions are grouped together under the ‘other religions’ name. The differences between the different religions within this group make this choice less than optimal, however for statistical reasons related to power the choice was unavoidable.
In their LinkedIn based study Zelekha et al. (2014) found a higher chance to be self-employed for Jews, compared to non-Jews. Researching the relationship between Judaism and self-employment, in a country where Judaism is not the main religion (USA and UK), using census data, marriage records and other sources of data ranging from 1800 until 1914, revealed an increased chance to be self-employed for Jews, compared to non-Jews (Godley, 2001). This positive relationship between Judaism and the chance to be self-employed can be explained by the finding of higher rates of return for investments made by Jews in human capital, such as the interaction with employees and suppliers, as well as better returns from years of schooling and labour market experience compared to non-Jews (Godley, 2001). This would imply that when participated in the same amount and level of education and work experiences a Jewish person would benefit more from these experiences compared to a not Jewish person. And for the investments in human capital higher results will be achieved by Jews then non Jewish persons. These higher rate of returns make the Jewish self-employed more capable and experienced while receiving a higher rate of return on human capital in the form of better relations with stakeholders and better motivated employees. These findings can potentially indicate that Jews have a higher chance to be self-employed, compared to non-Jews. It is also implied that Jews have a tendency to over-educate themselves and that this may also be a reason for the positive relationship with the amount of self-employment. This however has not been strongly supported by current research.
The relationship between Buddhism and the chance to be self-employed had conflicting results in the overarching quantitative studies on the relationships of different religions with the amount of self-employment (Audretsch et al., 2013; Zelekha et al., 2014). In a qualitative study on the influence of Buddhism on the entrepreneurial decision, conducted with interviews with Buddhists living in Canada and Buddhists living in Nepal, the outcome was unclear whether Buddhism has a positive or negative influence on the entrepreneurial motivation (Vallerie, 2008). However it was clear that there is a relationship between self-employment and the Buddhist religion. The values from the Buddhism’s religion seem to clash with some of the
10 traditional goals to be self-employed, mainly on the economic aspects. Balancing these tensions between the goals of the business and the values from Buddhism is key to successful self-employment for these Buddhists. The self-employed Buddhists interviewed for this case study also report choosing a livelihood that would avoid creating karma. The desire to not create karma is described as the reason for these Buddhists to choose self-employment. However with an alternate interpretation of not creating karma one could argue that Buddhists are motivated to not engage in self-employment. Because the findings of the different researches do not agree it is hard to predict the potential influence being Buddhist has on the chance to be self-employed, compared to non-Buddhists.
Lastly for the still many remaining different religions there no individual studies have been found on their relationship with the amount of self-employment. However there has been a study on the relationship of reported atheists or agnostics and the chance to be self-employed. In this empirical study by Wiseman (2013) on the differences of religious influences on entrepreneurship per state in the U.S. a positive relationship has been found between being atheists or agnostic and the chance to be self-employed. The relationship is being explained by the higher levels of self-interest and lower levels of costs due to religious behaviour of atheists and agnostics compared to people involved in a religion.
Ethnic Background
Ethnicity potentially influences the religion one belongs to, as the majority religion of the ethnic group is most likely taken with them when they immigrate. Therefore ethnicity is an important factor when discussing the relationship between self-employment and religion. Members of some non-white ethnic groups have entered self-employment in numbers greater than expected based on their presence. And because most minority religions in the UK are bound to one or more ethnic groups, ethnicity could be a confounding factor for the relationship between religion and the chance to be self-employed.
The UK, like many industrialised countries, has become considerably more ethnically diverse in recent decades. From 1991 to 2011 the foreign-born population of the UK rose by 4 million to 7.5 million people, or 13% of the population. These foreign-born UK residents have a different cultural background, often with a religion that is not the same as the UK’s main religion, Christianity.
Aldrich & Waldinger (1990) propose a framework for understanding the high number of non-white ethnic business owners. It is built on three components: Opportunity structures, Group characteristics and Ethnic strategies. Opportunity structure entails the opportunities for
11 potential non-white ethnic business owners. However the market conditions and the access to ownership are usually predetermined and controlled by white ethnic group members (Aldrich & Waldinger 1990). This could potentially suggests that non-white ethnic self-employed would have a harder time to secure the needed resources for self-employment. Thus the opportunity structure would suggest a smaller amount of non-white ethnic self-employed. However the component Opportunity structure does not explain necessity driven entrepreneurship. Necessity entrepreneurs are entrepreneurs that choose self-employment because they lack better work alternatives and/or have been unemployed for an extended period of time (Block & Koellinger, 2009). According to Hou & Wang (2011) immigrants report to have entered self-employment more often because they were unable to find a paying job in comparison to non-immigrants. On the other hand high skilled and well educated immigrants have a better perspective for paid employment, nevertheless highly skilled immigrants have the highest chance to be self-employed (Rodriguez-Pose & Hardy, 2014). When people from non-white ethnicity start a business it is more often a family business compared to the businesses started by white ethnic people. The non-white ethnic family businesses move down in the family and provide work for the complete family. This way the second generation, which are no longer immigrants but still have an non-white ethnic background, can be forced into self-employment (Basu, 2004). These reasons could potentially explain some part of the high numbers of non-white ethnic self-employment for low and high educate people from non-white ethnicity.
It has become clear from this chapter that there is some work on the relationship of different religions with the chance to be self-employment. However only one quantitative study tried to control for the ethnicity of the participants, yet they did this by constructing a database that does not describe the real population (Zelekha et al., 2014). Therefore none of the previous studies have results that are generalizable for populations outside their own sample. To create generalizable results for western countries the following research question is proposed:
Does religiosity relate to the tendency to be self-employed in the UK, after controlling for ethnicity?
12
Method
Data
The source of data used to find a relationship between religiosity and self-employment comes from the Understanding Society longitudinal study in the UK (University of Essex, 2008). The Institute for Social and Economic Research is the data provider and the UK Data Service is the data distributor for the data used in the Understanding Society study. The main survey study is used which has four waves released as of November 2014. Each wave consists of data collection over the period of 24 months, with overlap between them. The waves span from 2009 with wave one finishing in Q4 2010. Four more waves will be finished until 2017. From the four available waves the first wave is being used in this study. The reason for using the first wave and not the most recent wave, number four, is that the number of respondents on the first wave were the highest.
The data were mostly collected via face to face interviews in the respondent’s home, conducted by trained interviewers. The interviewers were instructed to assess the English ability of non-native English speakers. After assessing the English speaking ability the questions could be changed to better fit the English speaking ability of the respondent. In the case of no English speaking ability at all two different processes were used. When there were bi-lingual family members present they were asked to assist. In case no bi-lingual family members were present (or no family members were prepared to translate) an accredited bi-lingual interviewer would be assigned to the case. The interviews were on a yearly basis. Only household members from the age of 16 and up are interviewed. The study is funded by multiple government departments within the UK. There was a consent form in the information material the respondents received. The consent was asked to link administrative health and education records. Of course all respondents are coded to remain anonymous.
Sample
Initially 55,755 households were approached for participation in the study. From all of the contacted households 52.6% agreed to participate in the study. When households didn’t respond six phone calls were made to try and contact them, after which attempts to contact the households where ceased. This was the case in 15.1% of the potential households. The remaining 32.3% refused to be part of the study. The respondents were filtered on two criteria, being part of the working force (age 18 through 65) and secondly being either self-employed or an employee. The amount of money the respondents make, either as employee or
self-13 employed has not been regarded. Because the nature of the question: “How much are you usually paid?” is sensitive and therefore a the response rate on the question was very low. This resulted in a sample size of 27,279 respondents. The sample consisted largely of new respondent, with added respondents from an ethnic minority boost sample to ensure enough participation of ethnic minorities and north Irish participants. The demographic characteristics are displayed in table 2 and 3.
Study materials
To test the relationship between religion and the chance to be self-employed the results of various demographic questions from the Understanding Society questionnaire were selected. These questions measure whether or not a person is self-employed, what religion they belong to and their ethnicity. For self-employment the question used was: “Are you an employee or self-employed?” This question was only asked if it became clear that the respondent had a paying job. To determine the respondent’s religion two questions where used. First they were asked whether or not they regard themselves as belonging to any particular religion. When the respondents answered positively the follow up question was: “Which religion do you regard yourself as belonging to?” For the determination of the ethnic group the respondent belongs to the question: “What is your ethnic group?” was used. Other demographic measurements such as age, sex and level of education were also controlled for. All other respondents were taken as a part for this study, however to prevent any form of bias each case with a missing answer in any of the variables were completely filtered out for the study.
The questions used were open questions, however once answered the answers were analysed and categorized.
Statistical Power
A minimum sample size is needed for the test for the relationship between the different religions and the chance to be self-employed to enable statistical significant results. To test this, religions are treated as individual dummy variables to test for their relationship with the chance to be self-employed. Therefore the power to detect a relationship between the religious variable and self-employment for the various religions is calculated, assuming a medium effect size with an odds ratio of 1.2 and an R2 of 0.1. The alpha for the two sided test is known at a level of 0.05 as are the sample sizes for the different religions. XLSTAT 2015 software was used, and the results are displayed in table 1. As the software cannot compute power for samples that are smaller than 1% of the complete population, the values for Buddhism, Judaism and the other religions are not accurate. However since the smallest possible range (1%) gives only a power
14 of 0.277 it is safe to assume the even smaller religions will have an even smaller power. A minimum power of 0.8 is preferred to limit the chance of a type II error, resulting in a failure to reject the null hypothesis even though it is false. As is shown in table 1 only Christianity and the Islam have high enough power to meet the minimum 0.8 criteria. The other individual religions all have samples that are to small given the assumed R2 and odds ratio. However excluding all other religions from the research would not be desirable. Therefore a new variable is created called other religions, which contains the following religions: Hinduism, Sikhism, Buddhism, Judaism and other. The new variable other religions has a sample which allows for a high enough test power to be considered reliable.
Religion Sample size (percentage of
whole population) Power Christianity 9,336 (34.2%) 1.000 Islam 1,543 (5.7%) 0.873 Hinduism 560 (2.1%) 0.493 Sikhism 282 (1%) 0.277 Buddhism 132 (0.5%) <0.277 Judaism 67 (0.3%) <0.277 Other 252 (0.9%) <0.277
Other religions (new variable) 1,293 (4.7%) 0.809
Table 1. Power according to XLSTAT 2015, two-sided test with alpha = 0.05 and OR = 1.2
Dependent variable
The dependent variable is whether or not an individual is self-employed. This is compared to a control group that are employees. In the UK 14% of the population is self-employed (Office for National Statistics, 2011), while in this sample 12.9% is self-employed.
Independent variables
The independent variables are religiosity, measured as Christianity, Islam or other religions. Using dummy variables the scores on religiosity can either be 1 (belong to the religion) or 0 (do not belong to the religion). Within the Christianity variable all different sorts of Christianity will be combined (protestant, roman catholic, Baptist, Methodist, etc.). The religions combined in the other religions variable are Hinduism, Sikhism, Buddhism, Judaism and other answers that were not Christian or Muslim.
15
Control variables
The control variables are chosen for their probability to effect occupational choice. Individual characteristics highest level of education completed, age, marital status and sex have impact on self-employment decision (Block, 2012; Gupta, 2009; Kautonen, 2008; Parker, 2009) and are therefore used as control variables. Ethnic background is also controlled for to see if minority religions do not score high on numbers of self-employed because of their non-white ethnic background (Aldrich & Waldinger 1990). For the ethnicity the measure divides participants in two categories. Either the participants ethnicity is seen as white ethnicity (indicate by the white skin of the participants, or non-white ethnic (indicated by the non-white skin of the participants). Education is measured as the highest earned qualification with the use of dummy variables. The categories are: secondary education, meaning no more education after the secondary level was completed with results which do not enable entrance to a university, pre-university, meaning no more education after the secondary level was completed with results which do enable entrance to a university, higher education, meaning further education after the secondary level, not on university level, and university, meaning further education after secondary level on the university level. These variables are all coded into categorical variables, except for age. In the case of sex 0 = female and male = 1, for ethnicity white ethnic background = 0 and non-white ethnic background = 1, for marital status 0 = not married and 1 = married and for highest educational qualification 1 = secondary education, 2 = pre-university, 3 = higher education and 4 = university. Age is measured on a continuous scale.
Correlation
The correlations between all the used variables, including the dependent variable self-employment, a correlation analyses being used. This analysis shows the correlation of each variable on an individual level with the correlation of each other variable. This provides a good overview of the how the sample and how the variables are linked to one another.
Descriptives
Descriptives analyses are used to determine the descriptive properties of the variables. The analysis shows mean scores for variables, the amount of participants that fall into each group and a quick comparison to the UK average. This helps to see how the sample is divided and how closely it matches the UK population averages.
16
Model
The model estimated for this research is a Logistic regression model. It links the specific type of religion and the control variables to self-employment. The choice for logistic regression is made because the outcome is binary (someone either is completely self-employed or an employee). The analysis is built up in two blocks, first testing for the relationship that religions have with the chance to be self-employed, and in the second block the control variables: ethnicity, age, education and sex are added. The following models are used to estimate the chance of someone to be self-employed:
Step one: 𝑃(𝑆𝑒𝑙𝑓 − 𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑑) = 1/(1 + 𝑒 − (𝛼 + 𝛽1 ∗ 𝐶ℎ𝑟𝑖𝑠𝑡𝑖𝑎𝑛𝑖𝑡𝑦 + 𝛽2 ∗ 𝑀𝑢𝑠𝑙𝑖𝑚 + 𝛽3 ∗ 𝑂𝑡ℎ𝑒𝑟 𝑅𝑒𝑙𝑖𝑔𝑖𝑜𝑛 ) ) Step two: 𝑃(𝑆𝑒𝑙𝑓 − 𝑒𝑚𝑝𝑙𝑜𝑦𝑒𝑑) = 1/(1 + 𝑒 − (𝛼 + 𝛽1 ∗ 𝐶ℎ𝑟𝑖𝑠𝑡𝑖𝑎𝑛𝑖𝑡𝑦 + 𝛽2 ∗ 𝑀𝑢𝑠𝑙𝑖𝑚 + 𝛽3 ∗ 𝑂𝑡ℎ𝑒𝑟 𝑅𝑒𝑙𝑖𝑔𝑖𝑜𝑛 + 𝛽4 ∗ 𝐸𝑡ℎ𝑛𝑖𝑐 𝐵𝑎𝑐𝑘𝑔𝑟𝑜𝑢𝑛𝑑 + 𝛽5 ∗ 𝐴𝑔𝑒 + 𝛽6 ∗ 𝐴𝑔𝑒2 + 𝛽7 ∗ 𝑆𝑒𝑥 + 𝛽8 ∗ 𝐸𝑑𝑢𝑐𝑎𝑡𝑖𝑜𝑛 + 𝛽9 ∗ 𝑀𝑎𝑟𝑖𝑡𝑎𝑙 𝑆𝑡𝑎𝑡𝑢𝑠) )
17
Results
Table 2 shows the amount of self-employed and employees for every group. With this table it is easily found whether each group has a higher or lower amount of self-employed than the sample average of 12.9%. Reviewing this table shows that in the groups of the Muslims, the males and the group with no/other education the amount of self-employed is much higher than the sample average. Groups with low amounts of self-employment are the women and the unmarried.
Self-employed (percentage) Employee (percentage)
Christian 1,102 (11.8%) 8,234 (88.2%) Non-Christian 2,427 (13.5%) 15,516 (86.5%) Muslim 270 (17.5%) 1,273 (82.5%) Non-Muslim 3,259 (12.7%) 22,477 (87.3%) Other Religion 173 (12.9%) 1,120 (86.6%) No Other Religion 3,356 (12.9%) 22,630 (87.1%) Religious 1,649 (12.7%) 11,326 (87.3%) Not religious 1,558 (12.6%) 10,779 (87.4%) White Ethnicity 2,606 (12.7%) 17,862 (87.3%) Non-white Ethnicity 600 (12.4%) 4,240 (87.6%) Married 2,143 (14.7%) 12,417 (85.3%) Not Married 1,066 (9.9%) 9,690 (90.1%) Male 2,419 (17.9%) 11,062 (82.1%) Female 1,110 (8%) 12,688 (92%) Highest educational qualification: -University 1,017 (12.4%) 7,159 (87.6%) -Higher Education 357 (11.5%) 2,750 (88.5%) -Pre-university 313 (11.8%) 2,336 (88.2%) -Secondary Education 1,116 (12.4%) 7,901 (87.6%) -Other/no Education 762 (16.9%) 3,739 (83.1%)
Mean age (in years) 45 40
18 In table 3 the different demographic descriptives of the population are shown and compared to the average of the UK. This way the differences between the sample and the UK population can easily be spotted. Table 3 shows large differences between the amount of Christians in the sample and the amount of Christina in the UK. In the sample the amount of Christians is much lower than in the UK. For the not religious groups and the group that has university as highest educational qualification on the other hand the groups are much larger than the UK average.
Category Part of category Not part of category UK nationwide
Male 13,481 (49.4%) 13,798 (50.6%) 49.2% Self-employed 3,529 (12.9%) 23,750 (87.1%) 14% Christian 9,336 (34.2%) 17,943 (65.8%) 59% Muslim 1,543 (5.7%) 25,736 (94.3%) 5% Other religion 1,293 (4.7%) 25,986 (95.3%) 4% Not religious 12,337 (45.2%) 12,975 (47.6%) 25% Non-white Ethnicity 4,840 (17.7%) 20,468 (75%) 13% Married 14,560 (53.4%) 10,756 (39.4%) 47% Highest educational qualification: -University 8,176 (30%) 19,103 (70%) 13.6% -Higher Education 3,107 (11.4%) 24,172 (88.6%) 13.6% -Pre-university 2,649 (9.7%) 24,630 (90.3%) 12.3% -Secondary Education 9,017 (33.1%) 18,262 (66.9%) 28.6%
Table 3. Demographic descriptions of groups compared to UK averages
Correlation
Table 4 shows the correlation between the main variables in this study. The table shows a negative correlation between self-employment and Christianity, while for the Islam the correlation with self-employment is positive. Furthermore the other religions variable does not have significant correlation with self-employment. For correlation with self-employment the control variables non-white ethnicity and most levels of education (University, Pre-university and secondary education) do not have a significant result. However Higher education does show a significant correlation to being self-employed and is of a negative nature. Other control variables (age, age2, being married and being male) all have a significant and positive
19 The control variables’ correlations with the different religious variables differ slightly between them. While being married and secondary education as highest educational qualification correlate to all the religious variables in the same direction (positive and negative respectively), the other control variables show more variation. The most notable differences are with the correlations of age and sex with the religious variables. Age has a positive correlation with both Christianity and other religions and a negative correlation to the Islam, suggesting that as you get older the chance of being a Muslim gets smaller. For the correlation for being male the difference lies with Christianity and the Islam, as the correlation between sex and Christianity is negative while the correlation of sex and the Islam is positive. This indicates that being male gives a higher chance to belong to the Islam and a lower chance to belong to Christianity.
Ethnicity (measured as being from a non-white ethnic background) shows a strong positive correlation with both the Islam and other religions while the correlation between non-white ethnicity and Christianity is small and negative. In other words non-non-white ethnic people have a higher chance of being either a Muslim or part of another religion. The non-white ethnicity goes along with cultural differences from places where Christianity is not as regular, the white ethnic background culture goes along with cultural differences from places where Christianity is regular. For the Islam and other religions this effect is opposite.
20 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (1) Self-employed (2) Christian -0.024** (3) Muslim 0.033** -0.177** (4) Other religion 0.003 -0.161** -0.055** (5) Ethnicity (Non-white) -0.004 -0.075** 0.474** 0.356** (6) Marital Status 0.071** 0.098** 0.067** 0.043** 0.039** (7) University 0.010 0.008 0.043** 0.075** 0.125** 0.020** (8) Higher Education -0.015* 0.056** -0.021** -0.002 0.007 0.032** -0.220** (9) Pre University -0.011 -0.025** 0.022** 0.000 0.018** -0.076** -0.199** -0.095** (10) Sec- Education -0.012 -0.030** -0.064** -0.067** -0.138** -0.042** -0.450** -0.238** -0.215** (11) Age 0.133** 0.180** -0.121** 0.035** -0.146** 0.354** -0.074** 0.020** -0.132** -0.048** (12) Age2 0.130** 0.179** -0.122 -0.038** -0.153** 0.318** -0.088** 0.017** -0.116** -0.053** 0.989** (13) Sex 0.147** -0.142** 0.089** 0.012 0.060** 0.036** 0.004 -0.088** 0.002 0.005 0.020** 0.021**
Table 4. Correlation between variables * Correlation is significant at the 0.05 level (2-tailed) ** Correlation is significant at the 0.01 level (2-tailed)
21
Logistic regression
The results from the regression analysis are displayed in table 5. In the first step of the model only the variables for religions are present: Christianity, Islam and other religions. From the first step we can see that without the control variables in the model the odds ratio for Islam is significant (OR = 1.465 p < 0.000), the odds ratio for Christianity is marginally significant (OR = 0.923 p = 0.054) and the odds ratio for other religions is not significant.
The odds of the relationships indicate that a person being Christian has a 0.923 odds ratio to be employed compared than a not Christian. As for Muslims the odds to be self-employed is 1.465 odds ratio greater than for not Muslims. Because the other religions variable is not significant being part of another religion does not influence your chance to be self-employed compared to those who do not belong to another religion.
In the second step of the model the control variables were added. The effect size Christianity decreases and becomes statistically significant with the addition of the control variables. The effect size for the Islam increases with the addition of the control variables, while the significance stays equal. For the other religions the effect size increases and it becomes statistically significant with the addition of the control variables. In this model, being Muslim on the other hand increases the odds to be self-employed by 1.802 odds ratio, compared to non-Muslims. Being Christian decreases the odds ratio to be self-employed by 0.889 odds ratio compared to non-Christians. Lastly, being part of another religion (not Christian or Muslim) increases the odds ratio to be self-employed by 1.236 odds ratio compared to not being part of another religion, when other variables are being controlled for.
The relationship between ethnicity and the amount of self-employed is negative, therefore people from white ethnicity have a higher chance than to be self-employed than people from non-white ethnicity. The difference between people from white ethnicity and people from non-white ethnicity is 0.804 odds ratio. The control variable age has small effect sizes but it is measured on a continues scale and with each year older than eighteen the effect increases. The control variable gender has a relationship with self-employment that increases the odds to be self-employed, while being male, with 2.323 odds ratio compared to females. Furthermore the relationship between being married and the amount of self-employment is also positive, as married people are 1.112 odds ratio more likely to be self-employed than not married people are, when other variables are being controlled. Finally the relationships between the various levels of education (university, higher education, pre-university and secondary education) have bad significance as only secondary education has a marginally significant relationship with the amount of self-employed. Having completed secondary education as the highest educational
22 qualification decreases the odds to be self employed by 0.909 odds ratio compared to people who do not have secondary education as their highest level of educational qualification, although this result only shows marginal significance.
Regression
Variables Odds Ratio (S.E.) Model 1
Odds Ratio (S.E.) Model 2 Constant 0.145 (0.026)*** 0.006 (0.268)*** Christianity 0.923 (0.041)* 0.889 (0.095)*** Islam 1.465 (0.072)*** 1.802 (0.095)*** Other Religions 1.066 (0.086) 1.236 (0.100)** Non-white Ethnicity 0.804 (0.068)*** Marital Status 1.112 (0.044)** Highest educational qualification: -University 0.976 (0.057) -Higher Education 0.920 (0.074) -Pre-University 1.012 (0.079) -Secondary Education 0.909 (0.057)* Age 1.099 (0.013)*** Age2 0.999 (0.000)*** Sex 2.323 (0.041)***
23
Assessing the model
The multiple step approach of the analysis allowed for a comparison between the model with only the independent variables (Christianity, the Islam and other religions) after which the same analysis was conducted but with the various demographic control variables added. However to control the appropriateness and adequacy of the different models used in step one and two, and to be able to assess the improvement of the model over the steps, not only the improved significance can be assessed. The significance of each variable is calculated through the use of the Wald statistic, displayed in table 6. The increase in the Wald statistic is interpreted as an improvement in the significance of the variable associated with the Wald Statistic. For example the variable other religions had a Wald score of 0.558 in the first model and a significance of 0.455 that goes with it. However in the second model the Wald score had increased to 4.517 and the significance has improved to 0.034, making it possible to conclude what the relationship between other religions and the chance to be self-employed is.
Wald (Model 1) Sig. (Model 1) Wald (Model 2) Sig. (Model 2)
Christianity 3.720 .054 7.299 .007 Islam 28.176 .000 38.736 .000 Other religions 0.558 .455 4.517 .034 Non-white Ethnicity 10.268 .001 Marital Status 5.896 .015 Age 55.170 .000 Age2 23.832 .000 Sex 428.439 .000 Highest educational qualification: -University 0.175 .676 -Higher Edu 1.283 .257 -Pre-university 0.024 .877 -Secondary Edu 2.841 .092 Constant 5421.287 .000 352.907 .000
24 In addition to the improvement of the individual variables the usefulness of the models as a whole is also assessed. Nagelkerke’s R2 measurement is used to indicate the usefulness of the
independent variables for predicting the outcome of the dependent variable, making it the measure of effect size for the model. The Nagelkerke R2 is based on the -2 Log Likelihood (-2LL) of the model. The -2LL and Nagelkerke R2 are displayed in table 7. The decrease in the -2LL and the increase or the Nagelkerke R2 indicate that the model has improved. In other words, the independent variables of the first model explain only .7% of the predicted chance to be employed, while the second model explains 7.6% of the predicted chance to be self-employed. Although the model still is sub optimal, and will always be as there are many more factors influencing the chance to be self-employed, the model almost improved eleven times in its explanatory power. Indicating that the control variables added are improving the model.
Model 1 Model 2
-2LL 19,196.920 18,190.484
Nagelkerke R2 .003 .076
Table 7. Assessing the models fit
The variable ethnicity and the religious variables: Christianity, Islam and other religions might explain the same variance in the model (multi-collinearity) because whether someone has a white ethnic background or not could potentially influence their religion, making it more likely to be Christian for people with white ethnicity and more likely to be either Muslim or part of another religion with non-white ethnicity. To make sure this is not the case in this study the multi-collinearity was checked. The model showed that the religious variables explain 38.2% (R2 = 0.382) of the variance in the ethnicity variable (whether or not a person has a white ethnic background). This amount of collinearity should not cause problems for the analysis of this study.
Of the 27,279 remaining respondents 1,977 different respondents had missing values within their interviews. Of these 1,977 missing respondents 1,957 were missing values for the question: “What is your ethnic group?”. Since the missings are heavily clustered around this question a robustness check is done to compare groups, one with these missing values and one without. The results of the robustness check are shown in table 8. The most notable difference are the large amount of male respondents who are part of the missing group (1,492 males vs 485 females). The robustness of the model seems fine with respect to the other variables.
25 Self-employment 3,529 (12.9%) 3,204 (12.7%) Christianity 9,336 (34.2%) 9,325 (36.9%) Islam 1,543 (5.7%) 1,542 (6.1%) Other Religions 1,293 (4.7%) 1,293 (5.1%) Non-white Ethnicity 4,840 (17.7%) 4,837 (19.1%) Male 13,481 (49.4%) 11,989 (47.4%) Married 14,560 (53.4%) 14,543 (57.5%) Highest educational qualification: -University 8,176 (30%) 7,636 (30.2%) -Higher education 3,107 (11.4%) 2,957 (11.7%) -Pre-university 2,649 (9.7%) 2,400 (9.5%) -Secondary University 9,017 (33.1%) 8,277 (32.7%)
26
Discussion
Summary of results
From these results it can be learned that different religions have different relationships with self-employment within the UK. Although differences per religion were suggested by the literature, the particular relationships found between the religions and the chance to be self-employed were different than what was suggested by the literature. Furthermore the two biggest religions in the UK, Christianity and the Islam, both have a distinctively different relationship with the chance to be employed. Whereas Christians have a smaller chance to be self-employed, compared to non-Christians, Muslims have a greater chance to be self-self-employed, compared to non-Muslims. Furthermore the relationship between other religions and the amount of self-employment is also positive, although not as strong as it is for Muslims, indicating that being part of another religion increases the chance to be self-employed compared to people who are not part of another religion. The research question posed in this study was: Does religiosity relate to the tendency to be self-employed in the UK, after controlling for ethnicity? With the results from the analyses it is now possible to answer this question. Religiosity does relate to the tendency of to be self-employed within the UK, however the nature of this relationship is dependent on the type of religion as not every religion has a positive relationship with the tendency to be self-employed.
Interpretation of the results
Since the relationship of religiosity is not straight forward the individual relationships of the religious variables are further examined and explained here. Firstly, the negative relationship between being Christian and the chance to be self-employed is described. This negative relationship means that being Christian lowers the chance to be self-employed, compared to non-Christians, while other variables are being controlled.
A possible explanation for this phenomenon is related to the values that belong to being Christian. In many instances the Christian values has conservative influences on decision making. Therefore some Christians may interpret the religious values as a more strict set of rules or laws. Because of their conservative interpretation of the Christian values their tendency to be self-employed might be lowered. Strict upholding of the Christian values could potentially impede self-employment. This is in contradiction with the findings of Griebel et al. (2014) which indicate that the Christian entrepreneurs stopped wage working because of the inability to combine their work identity with their religious identity. This does not seem to be the case
27 here as 88.2% of the Christians are wage working. As for results of the study of Rietveld & van Burg (2014) which indicate that protestant Christians that are self-employed believe their work is a vocation of God more often than protestant Christians that are wage workers. For the results of this study this would mean that the Christians are less likely to believe their work is a vocation of God and therefore they are less likely to be self-employed. However these differences could be ascribed to the different interpretation of the Christian values between more devout Christians and the less devout Christians. The sampling methods used by Griebel et al. and Rietveld & van Burg might have given them a sample of more devout Christians than the sample of this study. Therefore it might be that more devout Christians have a higher chance to be self-employed, compared to non-Christians and less devout Christians. But as this study, as well as the studies looked at, do not measure for the devotion of the participants to their religion this cannot be supported by the current findings.
However previous qualitative findings indicate that being Christian has a positive relationship with the chance to be self-employed. To explain the differences in these findings compared to the results of the analyses the characteristics of the other studies need to be examined. As mentioned earlier the samples used by Audretsch et al. (2013) and Zelekha et al. (2014), who both reported positive relationships between the chance to be self-employed and Christianity, both do not resemble a generalizable population for the UK (Audretsch uses an Indian sample while Zelekha uses a self-constructed LinkedIn based sample). Whereas the sample used here is generated from a highly regarded UK based longitudinal study. This makes the results displayed here more likely to be applicable to other first world nations, especially those that have a similar population-structure. Furthermore the lack of control for ethnicity in Audretsch’s study, which potentially is an important influence for the minority religion that Christianity is in India, leads to the conclusion that their results do not apply to the population of the UK.
The relationship between being Muslim and the chance to be self-employed shows a higher chance to be self-employed when being a Muslim, compared to non-Muslims. After controlling for ethnicity (and the other control variables in the second model) the increase in chance to be self-employed for Muslims became even higher. The increase in the chance to be self-employed when being a Muslim was also found by Audretsch et al. (2013) and Zelekha et al. (2014) in their respective studies. Having non-white ethnicity itself showed a decrease in chance to be employed in the same model. But as Esser & Benschop (2009) described the Islamic self-employed seem to value their religious background more than their ethnic background.
28 Therefore the negative relationship of being of non-white ethnicity with the chance to be self-employed seems to be limited by the choice of the Muslims to adapt to the UK lifestyle. This way only the positive effects of their ethnic background (e.g. the niche markets) are being used.
As being of non-white ethnicity has a negative impact with the chance to be self-employed, compared to people from white ethnicity people, it indicates that non-white ethnic people have a higher chance of being employed than the white ethnic people. Therefore discrimination against people from non-white ethnicity cannot be supported by these results. However the relationship indicating a higher chance for Muslims to be self-employed, compared to non-Muslims could still be an indication of the discrimination. This discrimination could manifest as lower chances on the labour market for Muslims. Being unable to find a paying job could lead to the Muslims becoming necessity entrepreneurs. An important note here is that instead of being discriminated on their ethnic background the Muslims are discriminated on their religious background.
The positive relationship between the Islam and self-employment found also coincides with the findings in the literature focussed only on the Islam reviewed earlier. As described the values that belong to the Islam and that are described in the Qur’an and the sayings attributed to the Profit helps to improve the position of the Muslims with regard to becoming self-employed (Pistrui & Fahed-Sreih, 2010). The better conditions with regard to the start-up capital obtained (no interest on the loaned money), thanks to the Islamic banking system, might provide an advantage to Muslims compared to non-Muslims with similar skills and education. Businesses started by Muslims are more likely to be family businesses than those started by non-Muslims (Basu, 2004). Therefore the second generation owners of these family businesses are automatically self-employed, as family businesses enable or sometimes even slightly force people into self-employment.
Being from another religion (Hinduism, Buddhism, Sikhism, Judaism and others), also increases the chance to be self-employed, compared to people who are not part of another religion. In the studies by Audretsch et al. (2013) and Zelekha et al. (2014) the increase did not became evident, as their findings contradict each other. However this effect only became significant while the control variables were also added to the model, which Audretsch et al. (2013) did not do, nor did either Audretsch et al. (2013) or Zelekha et al. (2014) create the same variable as used here.
Similarly as the relationship between being part of another religion and self-employment in the first model, the correlation between other religions and self-self-employment
29 was not significant. Therefore conclusions about the relationship between other religions and self-employment can only be made while other variables are being controlled for. Because in the first model the negatively confounding control variables made it impossible to see the real effect of the other religions variable. With the control variables in place the increase in chance to be self-employed for members of other religions became clear. Because of the many religions that the other religions variable consists of it is hardtop explain the increase in chance for members of other religions, compared to non-members of other religions.
The control variable ethnicity (indicating non-white ethnicity) unexpectedly has a negative relationship with the chance to be self-employed for non-white ethnic people, compared to white ethnic people. This result is unexpected because according to literature on the relationship between non-white ethnicity and self-employment indicates higher chance to be self-employed when of non-white ethnicity. Even more so the religious variables that have a positive relationship with being self-employed (the Islam and other religions) both had high correlation with ethnicity. This indicates that the relationship between the religious variables and the chance to be self-employed is stronger than the relationship between ethnicity and the chance to be self-employed.
These results can help to explain how religiosity can increase economic performance. By focussing more on individual religions and their particular relationship with the chance to be self-employed the increase of economic development in relation with the members of that particular religion can partly be explained. The relationship between religion and the chance to be self-employed could potentially suggest that some part of the economic growth linked to religion is from an indirect effect. This effect is formed by the relationship between religion and the chance to be self-employed, which potentially could influences the economic growth.
Limitations
Although the sample used to study the relationship between various religions and the chance to be self-employed is large, the distribution of the sample was not ideal. As most of the sample was part of either the main religion in the UK (Christianity) or part of no religion, some variables did not possess the power needed. Therefore the variables lacking power were grouped to increase their sample size and as a result increase their test power. This resulted in the variable other religions which consists of Buddhism, Sikhism, Hinduism, Judaism and other. Although creating this variable was necessary given the available data the results should be considered with caution, as the various religions brought together within the other religions
30 variable do differ greatly from one another. The differences within the religions, e.g. the many different forms of Christianity, are not explored in this study. Same as with the other religions there remains a possibility that these different forms of Christianity are different from one another. The level of devotion from the participants was not measured for this study. However there might be interesting differences between the people who only belong to a religion and the very devout members who is actively involved in their religion.
Furthermore only 52.6% of the contacted households agreed to participate in the study, and of those there were still unusable respondents. This might be because the interview were very time consuming and therefore people that make long hours, either employees or self-employed, did not participate as much. Because the group that refused to participate is unknown it is hard to predict how they might have influenced the study. However with more respondents it might have been possible to use separate variables for more religions instead of the other religions variable. This way any possible difference between these religions may be found. The high amount of missing values on the question about the ethnicity of the respondents should also be noted. As the results for ethnicity were unexpected the missing values might have influenced this outcome.
Future research
For future research the relationship between the religions grouped within the other religions variable should be examined individually, for the UK. This will provide greater insight in which religions have a tendency to increase the chance to be self-employed and which religions have the tendency to decrease the chance to be self-employed.
Other interesting research would be to study the tendency of religion to impact economic performance within the UK. This macro overview enables more depth in other future research about religions and their relationship with economic performance or a subtopic of economic performance, like self-employment.
Lastly it is interesting to do similar research as this study with an added focus of the devotion of the participants. This would allow for more difference within groups from the same religion.
31
References
Aldrich, H. E., & Waldinger, R. (1990). Ethnicity and entrepreneurship. Annual review of sociology, 111-135.
Audretsch, D. B, & Keilbach, M. (2004). Entrepreneurship capital and economic performance.Regional studies,38(8), 949-959.
Audretsch, D. B., Bönte, W., & Tamvada, J. P. (2013). Religion, social class, and entrepreneurial choice.Journal of Business Venturing,28(6), 774-789.
Barro, R. J., & McCleary, R. (2003). Religion and economic growth (No. w9682). National Bureau of Economic Research.
Basu, A. (2004). Entrepreneurial aspirations among family business owners: an analysis of ethnic business owners in the UK. International Journal of Entrepreneurial Behavior & Research,10(1/2), 12-33.
Benjamin, D. J., Choi, J. J., & Fisher, G. W. (2010). Religious identity and economic behavior(No. w15925). National Bureau of Economic Research.
Bettendorf, L., & Dijkgraaf, E. (2010). Religion and income: Heterogeneity between countries.Journal of Economic Behavior & Organization,74(1), 12-29.
Block, J., & Koellinger, P. (2009). I can't get no satisfaction—Necessity entrepreneurship and procedural utility.Kyklos,62(2), 191-209.
Block, J. H., Hoogerheide, L., & Thurik, R. (2012). Are education and entrepreneurial income endogenous? A Bayesian analysis.Entrepreneurship Research Journal,2(3).
Carree, M. A., & Thurik, A. R. (2003). The impact of entrepreneurship on economic growth. InHandbook of entrepreneurship research(pp. 437-471). Springer US.
Chiswick, B. R. (1983). The earnings and human capital of American Jews.Journal of Human Resources, 313-336.
Dana, L. P. (2009). Religion as an explanatory variable for entrepreneurship. The international journal of entrepreneurship and innovation,10(2), 87-99.
Dodd, S. D., & Gotsis, G. (2007). The interrelationships between entrepreneurship and religion.The international journal of entrepreneurship and innovation,8(2), 93-104.
32 Essers, C., & Benschop, Y. (2009). Muslim businesswomen doing boundary work: The negotiation of Islam, gender and ethnicity within entrepreneurial contexts. Human Relations,62(3), 403-423.
Godley, A. (2001). Jewish immigrant entrepreneurship in New York and London 1880-1914: Enterprise and culture. University of Illinois at Urbana-Champaign's Academy for Entrepreneurial Leadership Historical Research Reference in Entrepreneurship.
Griebel, J. M., Park, J. Z., & Neubert, M. J. (2014). Faith and Work: An Exploratory Study of Religious Entrepreneurs. Religions, 5(3), 780-800.
Guiso, L., Sapienza, P., & Zingales, L. (2003). People's opium? Religion and economic attitudes.Journal of monetary economics,50(1), 225-282.
Gupta, V. K., Turban, D. B., Wasti, S. A., & Sikdar, A. (2009). The role of gender stereotypes in perceptions of entrepreneurs and intentions to become an entrepreneur.Entrepreneurship theory and practice,33(2), 397-417.
Hou, F., & Wang, S. (2011). Diverse entrepreneurs: why do immigrants enter and stay in self-employment?.Perspectives on Labour and Income,23(3), 5-16.
Iannaccone, L. R. (1998). Introduction to the Economics of Religion. Journal of economic literature, 1465-1495.
Kautonen, T. (2008). Understanding the older entrepreneur: Comparing third age and prime age entrepreneurs in Finland. International Journal of Business Science and Applied Management,3(3), 3-13.
Keown, D. (1996). Karma, character, and consequentialism.The Journal of Religious Ethics, 329-350.
Lehrer, E. L. (2004). Religion as a determinant of economic and demographic behavior in the United States.Population and development review,30(4), 707-726.
Lewis, B. E., & Churchill, B. E. (2008).Islam: The religion and the people. Pearson Prentice Hall.
Office for National Statistics, 2011 Census: Aggregate data (England and Wales) [computer file]. UK Data Service Census Support. Downloaded from: http://infuse.mimas.ac.uk. This
