This paper relates curiosity to economic development through its impact on human capital formation and technological advancement in pre-modern times. More specifically, we propose that exposure to inexplicable phenomena prompts curiosity and thinking in an attempt to comprehend these mysteries, thus raising human capital and technology, and, ultimately, fostering growth.

We focus on solar eclipses as one particular trigger of curiosity and empirically establish a robust relationship between their number and several proxies of economic prosperity.

We also offer evidence compatible with the human capital and technological increases we postulate, finding a more intricate thinking process and more developed technology among societies more exposed to solar eclipses. Among other factors, we study the development of written language, the playing of strategy games and the accuracy of folkloric explanations for eclipses, as well as the number of tasks undertaken in a society, their relative complexity and broad technological indicators.

Lastly, we document rising curiosity both at the social and individual levels: societies incorporate more terms related to curiosity and eclipses in their folklore, and people who observed a total solar eclipse during their childhood were more likely to have entered a scientific occupation.

…To document and elucidate this relationship, we combine different datasets. Proceeding in stages, we first show the direct impact of total solar eclipses on economic development.¹² We then investigate the core mechanism associating solar eclipses and human capital. We further document a more advanced technology in places where total solar eclipses are more prevalent. Lastly, we close the loop, confirming that solar eclipses stimulate curiosity. To lend additional credence to our results, we assess the scope of religion as an alternative channel, finding no supporting evidence relating curiosity to religion. A set of placebo tests also confines the impact of solar eclipses to curiosity.

…The most prominent compilation regarding the living modes of ethnic groups is the Ethnographic Atlas of Murdock1967. This work samples more than 1,000 ethnicities scattered around the globe and contains details about subsistence, marital practices and labour division.

…According to our hypothesis that inexplicable events, such as solar eclipses, can increase curiosity, we expected that individuals who were exposed to solar eclipses during childhood would be more likely to pursue careers in science. This is because scientific occupations involve inquiring about the workings of the world and require a curious mindset. In contrast, religious occupations are not embedded with this inquisitorial dimension, and experiencing an eclipse should not alter the probability with which these occupations are entered. Thus, for each individual, we create a dummy variable stating whether or not a total solar eclipse was visible from her birthplace city between the ages of 5–15. To determine local eclipse visibility, we compared the paths of totality of eclipses to the coordinates of each individual’s birthplace and only considered the eclipses that occurred during the relevant dates for each individual. This procedure assumes that individuals did not migrate far from their place of birth, as the visibility of solar eclipses extends about 100 km to the north and south of the path of totality. [For example] Johannes Browallius was born in Västrås, Sweden, which was within the path of totality of the total solar eclipse that took place on May 3, 1715, when he was 8 years old.

In addition to creating a dummy variable for solar eclipses, we also created a similar variable for volcanic eruptions to gather information on another potential source of curiosity. Because cities are often located some distance from volcanoes, we accounted for the fact that volcanic effects such as ash falls and smoke columns can be observed up to 100 km away. Therefore, we considered an individual to have been exposed to a volcanic eruption if their birthplace city was within 100 km of a volcano that experienced an eruption during their childhood.

…Our data show that a total of 5,635 total and annular solar eclipses occurred in the 3,500 years considered, with an average of 1.61 per year. However, it is necessary to consider that a solar eclipse is only visible from certain points on the globe, and, therefore, the frequency with which they occur at a given location decreases rapidly as precision increases. Thus, if we consider 100×100 km² cells,⁴⁵ solar eclipse frequency is reduced to only 1.17 solar eclipses per century. At the city level, Steel2001 (pg31) reported an average inter-eclipse span of 410 years. Considering these characteristics, at a given location, total solar eclipses are an almost random variable, albeit they are slightly more likely in the northern hemisphere.

…Other measures of human capital similarly increase with the number of eclipses. Most notably, Column (3) indicates a better understanding of solar eclipses as the phenomenon becomes more prevalent. This is a relevant result that speaks directly to our theory. We posit that people devote mental resources to the understanding of eclipses. Thus, more refined theories closer to reality should ensue from more intense pondering. Column (3) indicates that this is indeed the case: a 1% increase in the number of solar eclipses increases by % the probability of having an elaborate explanation for the phenomenon. Similarly, if we boost the number of eclipses from the 10^th → 90^th percentile of the distribution for all ethnic groups, on average, the probability of having no explanation for the phenomenon decreases from 69% → 56%.

Columns (4) & (8) indicate a greater abundance of words related to the concept of a Calendar in folktales and a more intense development of actual calendars among societies more exposed to eclipses.⁶⁰ Related to the idea that eclipses stimulate curiosity and quicken thinking, Column (5) indicates a higher frequency of terms related to Thinking when the number of eclipses increases, with an associated marginal effect of 0.24 more tales on this topic for each additional eclipse. Lastly, Column (9) clearly indicates that geometry tends to develop more as solar eclipses increase, which is in line with the idea that geometry is a useful tool for predicting eclipses.

We first note that ethnic groups incorporate about 1.09 more concepts related to this phenomenon into their tales when the number of events they witness increases by 1% (the sample average is 4.63⁠). Likewise, shifting ethnic groups from the 10^th to the 90^th centile of the eclipse distribution increases the number of predicted mentions of eclipses in folklore 4.05 → 5.38⁠. This result lends credence to our assumption that total solar eclipses are a relevant phenomenon; otherwise, they would not feature as part of folktales. The results in Column (2) are similar: a 1% increase in the number of eclipses increases the number of words related to curiosity in folkloric tales by 0.11⁠. Although this may seem negligible, in general, curiosity-related words are not common in folkloric tales, averaging only 0.86 at the society level. Once this is considered, the marginal effect becomes much more relevant, representing around 13% of the sample average.

Columns (3) & (4) focus on the individual-level effect of eclipses in terms of curiosity. These columns indicate that individuals who spotted an eclipse between the ages of 5 and 10 are about 7.19% and 3.96% more likely to become scientists in adulthood. As a reference, 8.62% of sampled individuals followed a scientific career. Although surprising, this result is compatible with our hypothesis: people exposed to an inexplicable event become more curious and have a desire to understand. Related, one can reasonably expect a lingering impact of solar eclipses on curiosity as the involved mechanics become better understood. To test this possibility, we run unweighted regressions comprising all individuals born before a given time, repeating the process for each century. Because the number of individuals in our database increases with time, doing so places more importance on the most recent cohort entering the regression. Thus, this procedure enables us to study the temporal evolution of the impact of solar eclipses on scientific occupations.⁶⁷ Figure 3 illustrates the results, indicating that, as time advances, the probability of a child who saw a total eclipse choosing a scientific occupation converges towards its average.⁶⁸ Nevertheless, solar eclipses remain meaningful until around the 7^th century, and there is a small resurgence of their importance during the Renaissance period.⁶⁹

…3.1. Lunar Eclipses In line with our hypothesis, the previous results support the idea that inexplicable (harmless) phenomena can promote growth through curiosity. Similar effects may be reasonably expected of lunar eclipses, despite these being less impressive and more elusive as they happen at night. Online Appendix Tables C.1–C.5 repeat the main regressions of the paper, introducing lunar eclipses. In each case, panel A replaces total solar eclipses for total lunar eclipses, while panel B incorporates both types of eclipse to compare the relative strength of each.

In general, replacing solar eclipses by lunar eclipses provides results very much aligned and compatible with our hypothesis. Except for cases using the Seshat database, the new results corroborate the association between eclipses and development and validate that human capital and technology are plausible avenues through which the aforementioned relationship could be operating. The relative failure of the Seshat data to reproduce the previous results is not as critical as it may seem, as most polity borders do not change or do so only slightly. Because lunar eclipses are visible from half of the Earth, a relatively constant area implies a relatively constant number of lunar eclipses, which are absorbed by the polity fixed effects.⁷⁰ However, this drawback becomes a virtue when using the Ethnographic Atlas because, indirectly, the introduction of lunar eclipses partly controls for area, thus reassuring us that the results using cross-sectional data are not due to size differences.

Similarly, the apparent inability of lunar eclipses to improve the understanding of the phenomena (Column 3 of Online Appendix Table C.3) requires some clarification. Indeed, a careful reading of the folklore data turns a potentially devastating result into a coherent one. First, the process resulting in an eclipse differs between solar and lunar eclipses. In the first, the Moon clearly blocks the Sun, and thus both bodies interact. However, the Sun does not directly intervene in lunar eclipses: these happen at night. Second, folklore data record tales wherein eclipses result from the interaction between the Sun and the Moon, for instance, because they are mating. Hence, lunar eclipses cannot enter this body of oral tradition, and this is captured by the non-statistically-significant coefficient.

Generally, when we simultaneously analyse the two types of eclipses, the results restore the importance of solar eclipses to the detriment of lunar ones. As previously mentioned, this result follows from the characteristics of lunar eclipses, which are less impressive and noticeable. In this sense, a strong effect arising only from solar eclipses is compatible with our theory. In summary, as demonstrated by the positive and statistically-significant coefficients associated with lunar eclipses, inexplicable events are likely to promote economic development. Moreover, this set of results favours the idea that the more mysterious an event is, the more it promotes thinking.

See Also: