Slavery in Korea, documented since the Three Kingdoms period (BC 57-AD 668), differed significantly from African slavery. Initially involving prisoners of war and criminals, it evolved into a privately-owned system. Korean society was stratified into four classes: Yangban (elites), Chungin (middle class), Sangmin (commoners), and Nobi (slaves). Only commoners bore tax and military burdens. The *Jongcheon* law (1430-1731) stipulated that a child was a slave if either parent was, ensuring a consistent slave population. However, the *Jongyang* law of 1731 changed this, declaring a child a slave only if the mother was a slave. This led to a dramatic decrease in the slave population, a shift likely driven by political conflicts between King YoungJo and powerful elite groups who relied heavily on slave labor. The ensuing societal instability led to increased slave desertions, as slaves sought freedom by fleeing into the informal economy. This study uses data from three counties in Kyungsang province to analyze this transition, focusing on the causal link between the 1731 regulation change, increased desertions, rising monitoring costs, and the eventual decline of slavery in Korea. The research posits that the increased desertion rates and the associated monitoring costs led to a decline in the demand for slaves, and subsequent replacement by wage laborers.

Existing research on Korean slavery primarily focuses on the legal system and demographic aspects of the slave population, with limited exploration of the economic evolution of slavery. Studies like those by Pyung (1982), Park (1986, 2007), Jeon (1989, 1998), Choi (1978), and Ji (1995) detail policy changes and the legal system. Other works, including Palais (1996) and Patterson (1982), examine the demographics of the slave population. However, studies on slave transactions and their economic efficiency are scarce, with notable exceptions being Chong (1983) and Lee (1981). This study aims to address this gap by analyzing the role of slave desertion and its consequences on the agricultural economy, a topic largely neglected in previous research. The study leverages social network theory to explain why slaves fled to the informal sector, highlighting the lack of social capital and trust within these sectors, leading to market fragmentation and high interest rates. The study also considers existing economic debates surrounding the efficiency of slavery in various contexts.

This study uses a combination of quantitative and qualitative data to explore the relationship between the 1731 regulation change, slave desertion, and the decline of slavery in Korea. The quantitative analysis relies on data from government registries of household tallies from three counties (Danseong, Eonyang, and Daegu) in Kyungsang province. These registries recorded household members, including slaves, and detailed information on slave departures (including desertions), arrivals, and changes in slave status. The study also utilizes data on 634 private slave transactions from various historical records and trading documents spanning 1690-1893. These data provide insights into slave prices and market dynamics. The methodology uses panel two-stage least squares (2SLS) analysis to examine the relationship between slave desertion, monitoring costs (proxied by real slave prices), slave population, and the 1731 regulation change. Real slave prices are calculated by deflating nominal prices using rice prices as the main representative good in the Korean agricultural economy. The study uses rice consumption data to contextualize real slave prices. Qualitative data, such as information from the official household tally records and old private records from elite groups, helps to understand the social context and reasons behind slave desertion. The study controls for potential endogeneity issues in the slave population by employing instrumental variables (IVs) in the 2SLS analysis.

The study's descriptive statistics reveal a significant increase in slave desertion rates following the 1731 regulation change. The Daegu-Choi family data shows desertion rates rising from 12.5% in 1681 to 56.1% in 1801, with an average of 20.8% over the period. Similar trends are observed in Danseong and Eonyang counties. The analysis of slave transactions shows a continuous decline in real slave prices after 1731, indicating reduced demand for slaves. The average real slave price was 708 kg of rice, equivalent to the annual consumption of approximately seven adult men. The 2SLS estimation results show a significant negative relationship between slave desertion and real slave prices, suggesting that higher desertion rates increased monitoring costs and lowered slave demand. The effect of the 1731 regulation change is also significant in the model. The ratio of slaves to wage workers decreased significantly after 1731 in Eonyang county, further indicating a shift towards wage labor. Data shows that slaves between 16-40 years of age, the most physically fit, comprised a significant portion of the agricultural workforce, illustrating slaves' role in agricultural production.

The findings strongly support the hypothesis that the 1731 regulation change, by altering children's slave status, increased slave desertion rates, leading to higher monitoring costs and lower real slave prices. The decrease in slave demand and the consequent shift towards wage labor represent a Pareto improvement, moving from an unstable slavery equilibrium to a more stable one where the high risk of escape reduced slave productivity. The informal nature of this transition, driven by individual slave choices and market dynamics rather than large-scale abolitionist movements, is a notable feature distinguishing Korean slavery's end from that in other global contexts. The economic consequences of this shift included a contraction of the agricultural economy due to the collapse of the slavery system. The study's insights are relevant to the broader literature on institutional change, particularly in understanding how informal processes can lead to significant societal transformations.

This study demonstrates that the decline of slavery in 18th and 19th-century Korea was an organic process primarily driven by slave desertions following the 1731 regulation change. This resulted in increased monitoring costs for slave owners, a decrease in the demand for slaves, and a consequent shift toward wage labor. The findings contribute to the understanding of institutional change, highlighting the importance of informal processes and the economic consequences of shifting from an unstable slavery system to a wage-labor system. Future research could explore the long-term impacts of this transition on Korean economic development and social structure.

The study acknowledges limitations concerning data availability. While the study utilizes a comprehensive dataset of household tallies and slave transactions, the data on slave transactions are limited compared to those available for other historical contexts like the antebellum South. The reliance on data from Kyungsang province might limit the generalizability of the findings to other regions of Korea. The proxy used for monitoring costs (real slave prices) might not perfectly capture the full complexity of monitoring costs. Further research using more detailed data from various regions and a broader range of variables could help to strengthen the findings and address potential biases.