Age patterns of migration among Korean adults in early 20th-century Seoul

Abstract

In this study, we examine age patterns of migration among adults who resided in Seoul in the early twentieth century. We use information, obtained from the Seoul household registers, on the length of time these adults lived in their current residences to estimate age-specific migration rates and construct migration life tables. Our findings point to the following: First, Seoul residents were quite mobile. On average, during the early twentieth century, Seoul residents moved approximately four times between their primary working ages of 15 and 64. Second, upper-class individuals were more mobile than lower-class individuals. While the upper-class individuals moved approximately five times between ages 15 and 65, the lower-class individuals moved less than three times. This class differential can be explained because Korean bureaucrats experienced frequent duty changes during this period. Third, household composition also affected migration rates. We compare our results with European migration studies and discuss the implications of these finding on urbanization that occurred in early twentieth century Seoul.

Keywords::

• the Seoul household registers
• migration life table
• class differentials in migration

Acknowledgements

An earlier version of this paper was presented to the 2012 Social Science History Association Annual Meeting in Vancouver, B.C. in November 2012. We would like to thank an anonymous reviewer, as well as Martin Dribe and Cameron Campbell, for their valuable comments and suggestions. We also appreciate research assistance by JuKyeong Lee. The digitization of Seoul registers was supported by the National Research Foundation of Korea (NRF-2010-32A-D00035). This work was supported by the new faculty research program 2012 of Kookmin University in Korea.

Notes

^1. A simulation revealed the critical contributions of these compositional effects: A natural increase in the mid-nineteenth century English cities would have been 5.82 per thousand without changes in age-sex distribution that resulted from migration. This would have been much lower than the actual increase, which was 11.1 per thousand (de Vries, 1990).

^2. Urban–rural fertility differential is another demographic factor that might contribute to urban growth. Because scholars have determined that urban fertility tended to be lower than rural fertility, urban–rural fertility differentials worked in opposition to urban growth. However, differences between the two factors were not substantial (Sharlin, 1986). Thus, we do not discuss differential fertility in detail.

^3. The characteristics of the Gwanmu Reform have been widely debated by Korean historians. While Shin (1978) argued that it was, in the most basic sense, a counter-revolution led by the Royal Court and the highest bureaucrats to subdue reforms requested from below, Kang (1978) contended that it was an internal attempt to modernize the Chosun Dynasty. The debate is still ongoing. Please see Oh (2007) for a recent review of the literature.

^4. Kim (1982) includes households that were missing the duration variable by assuming that they did not change residences during the heads of households' lifetimes. However, we are skeptical of this assumption. We considered those heads of households missing in our duration variable.

^5. If our data had covered the entire area, we would simply have compared the total number of households between years and estimated population growth rates. However, because available data were incomplete, we had to use an indirect method to estimate population growth.

^6. This assumption is reasonable in a pre-transition society such as early twentieth century Seoul.

^7. Strictly speaking, we also need to assume that in-migrants and external migrants in each age group were equal in size because we are estimating age-specific migration rates. We are unable to provide any strong evidence for or against this assumption. However, to support our approach, the data reveals that the age distribution of internal migrants was similar to the age distribution of in-migrants (results are not shown). This means that the age-profiles of these two groups were similar to one another. Although this information is incomplete, it also suggests that age-profiles of external migrants would, in all likelihood, be similar to these two groups. This result provides indirect evidence for the assumption that the size of in-migrants was equal to the size of external migrants in each age group.

^8. According to the model of age patterns of migration developed by Rogers et al. (2011), migration rates are high among young children and young adults. High migration among young adults is attributable to job-related movements. High migration among young children is consistent with frequent job-related parental movements. According to this model, another migration peak is the ‘retirement peak’ that may reflect the propensity of individuals to return to their hometowns or to move to places friendly to the elderly after retirement. In our analysis, we primarily focused on the working-age population (ages 15 through 64). Therefore, we did not examine these pre-labor force and post-retirement peaks.

^9. Dribe and Lundh (2005) showed that the TMR for Swedish male servants in nineteenth century rural areas was 20. This figure is much higher than our estimates. This result may suggest that early twentieth-century Seoul residents were less mobile than Swedish rural servants during the nineteenth century. However, our estimates are not comparable with Dribe and Lundh's (2005) estimates. While we define migration as movements made to different villages (dongs), the Swedish study defined migration as any changes in residence.