By Lars Harhoff Andersen and Jeanet Sinding Bentzen
One of the fathers of the social sciences, Max Weber, taught us that religion might have facilitated the transition towards modern growth and prosperity. The Protestant ethic supposedly spurred the development of capitalism by making hard work a deed cherished by God. In addition, monks were among some of the great inventors in history and many early universities emerged near monasteries. This has led many to conclude that Christianity has been a blessing for the economy.
Another literature has emphasized that religion may hamper economic growth for two main reasons. First, religion relegates time and resources to religious activities, thus reducing the resources available for productivity-enhancing activities (Barro and McCleary 2003, Campante and Yanagizawa-Drott 2015). Second, as most religions often involve a fixed worldview established long before the emergence of modern science, scientific discoveries have often been in conflict with religion (e.g., Chaney 2016, Benabou et al 2021). A famous example is the inquisition of Galileo Galilei by the Catholic Church for his championing of Copernican heliocentrism (seen in the painting above). Whether religion facilitates or impedes economic growth thus remains an empirical matter.
In recent work, we set out to identify the impact of religiosity on the advancement of science and economic growth in Europe during the past 700 years. Our results do not support the view that Christianity benefited economic growth.
Let us first briefly touch upon the difference between religiosity and types of religions. Max Weber conjectured a difference between Protestants and Catholics, which involves potential differences due to the type of religion. A literature has scrutinized the empirical support for Weber’s idea and other hypotheses involving differences between Muslims, Christians, Jews, Hindus, and others (reviewed in Becker et al 2016, 2021, and Kuran 2018). Another literature examines instead differences in terms of religiosity, or the degree of believing, independent of the god or gods to which the followers are devoted (e.g., Squicciarini 2020, Benabou et al 2021, review by Bentzen 2021). The idea is that the major religions involve stark similarities, and thus the general intensity of believing matters for societal differences. This research investigates the impact of differences in religiosity within Christianity, Islam, Buddhism, etc. Our research belongs with this latter strand.
To measure religiosity in a time without surveys, we exploit that our given names reveal facts about the identity and preferences of our parents, including their religious identity. When naming our children, we – consciously or unconsciously – tend to choose names that match our identity. Parents for whom religion plays a central role in life are potentially more likely to select a name for their children that is inspired by religion, compared to parents for whom other factors, such as ethnicity or nationality, are more important. We use given names to construct a proxy for average religiosity in two large subsets of similar individuals: 450,000 European-based authors born between 1300 and 1940 (shown in panel a of the figure below) and 50,000 university students born in the Holy Roman Empire in 1300-1550 (in panel b). The datasets are attractive as they provide information on first names, birthplace year and geocodes, and professions of individuals spread across time and space in historical Europe. We define a name as signaling Catholic religiosity if a major saint shared it and Protestant religiosity if it was shared by a major biblical figure (as many biblical figures were also sanctified, the two overlap a great deal). We hypothesize that religion played a larger role in the lives of parents who gave their child such a name, compared to parents who chose other names for their children.
We first set out to test whether our names-based religiosity measure does in fact capture differences in religiosity well. It turns out that individuals carrying names shared by significant religious figures are indeed more likely to engage in behaviors associated with religion. They are more likely to study theology or church law, and more likely to be born in areas ranking as more religious based on another measure of historical religiosity. This measure – whether the clergy in French districts revealed loyalty towards the Catholic Church in 1791 – is the only other proxy for historical religiosity that we know of. Last, we find that parents were more likely to name their children after religious figures when an earthquake had recently struck near their locality (see the figure below). The latter is a manifestation of religious coping – religious individuals tend to use their religion for emotional comfort when faced with adversity (e.g., Pargament 2001, Bentzen 2019, 2021).
The effects are quite large, but yet meaningful. Individuals who shared name with a major religious figure were 12% more likely to study theology, 2% more likely to study canon law, 10% more likely to be born in areas loyal to the church, and 8% more likely to be born just after an earthquake, compared to individuals whose name was not shared by a significant religious figure.
We next examine the impact of religiosity on science, innovation, and the transition to modern growth. Exploiting information on the professions of the authors, we find that those with religious names were more likely to become priests and theologians, but less likely to become scientists, engineers, and doctors (see the figure below). Exploiting information on study choices of the university students we also find that students with religious names were less likely to proceed with advanced studies, compared to the students without religious names.
We interpret these results as reflecting a causal impact leading from religiosity to a lower focus on science. While a competing hypothesis that the development of science diluted beliefs in God may simultaneously be true, this alternative direction of causality cannot explain our results. Profession or study choices later in life typically do not influence peoples’ given names. We can rule out that many alternative confounders explain the results. First, the samples consist of unusually similar individuals (either university students or authors), which means that factors such as social status are held fixed at rather similar levels. Second, the detail of the data allows us to compare individuals born in the same year, in the same locality, of similar gender, nobility status, ethnicity, and with names that are equally normal in general.
Even for these highly similar individuals, the intensity of religion in their upbringing had profound impacts on peoples’ choices later in life. Religious individuals were more likely to choose professions involving religion, while less religious individuals were more likely to choose professions relating to science. While there are certainly examples of clashes between religion and science throughout history (the trial of Galileo by the Catholic Church is one), there need not be such a conflict in order for the results to emerge. It may simply be that an elevated interest in religion crowds out potential time spent on other interests, such as science and engineering.
A natural next question then is how these differences in terms of religiosity mattered for the transition to modern economic growth. To measure economic growth across European cities, we use urbanization rates as is custom in the literature. We calculate average religiosity for each city by the share of authors with religious names born within a 100 km radius of the city center (panel a of the below figure shows the spread of these cities across Europe). Of course, religiosity of authors may not reflect the religiosity of the general population. However, as long as our measure allows us to rank the religiosity of cities quite well, the measure is useful for empirical analysis. Furthermore, the focus on one population group is methodologically attractive, as other things such as social status or education are rather similar within this particular population group. With these data, we document that cities inhabited by more religious individuals grew slower, compared to their less religious counterparts did. Panel b of the figure below shows average city size in two equally-sized groups of cities split by their 17th century religiosity level and serves as a simple illustration of this finding.
This negative impact of religiosity rises over time, particularly after the mid-1800s, coinciding with the second industrial revolution. This period, also known as the technological revolution, was a phase of accelerating economic growth due to rapid scientific discovery. Our results indicate that cities in which more time had been devoted to religious activities and less towards scientific discovery experienced slower economic growth. The reason being that scientific discovery became important for economic growth at this time. In the early middle ages, the opposite may have occurred, as monasteries potentially gave monks more time to engage in inventions, compared to full-time farmers or merchants. Religion may well have been good for growth in earlier times and may continue to exert positive effects on certain dimensions of the economy, such as reduced crime rates. However, our research shows that the negative effects dominate the positive in terms of aggregate economic growth rates.