Real estate has been widely recognized as an asset to mitigate inflation risk and has established itself globally as an attractive investment asset. The inflation hedging property of real estate refers to the characteristics whereby the cash flows gen...
Real estate has been widely recognized as an asset to mitigate inflation risk and has established itself globally as an attractive investment asset. The inflation hedging property of real estate refers to the characteristics whereby the cash flows generated from real estate increase in line with or beyond general inflation rate. Since the seminal study by Fama and Schwert (1977), extensive empirical research has examined the relationship between inflation and investment returns across various asset classes. While the inflation hedging properties of housing assets have been confirmed in numerous international studies, existing domestic research has often reported inverse or weak hedging effects.
The cash flows generated from owning real estate generally consist of capital gains and rental income. When these sources of income move in tandem with inflation, real estate can function as a hedge against future price increases. If rents can be flexibly adjusted in line with inflation, the hedging effect would be further enhanced. On the other hand, if rental contracts are adjusted passively based on past inflation, this hedging effect would be weakened. Accordingly, the effectiveness of real estate as an inflation hedge depends on the frequency of lease negotiation and the accuracy of inflation forecasts.
Futhermore, if housing asset returns across regions respond differently to the same inflationary shock, it would imply not only potentially widening regional asset disparities across regions but also that such differences go beyond the economic polarisation and may lead to comprehensive forms of social inequality across education, human capital, social capital and cultural capital. Particularly, when wealth polarisation is rooted in non-labour income, there is a risk that this entrenches economic hierarchies that cannot be resolved naturally through economic recovery alone.
This study investigates the inflation hedging properties of housing assets in South Korea, particualy focusing on apartments in Seoul, the country’s representative housing market. It evaluates not only its capital returns but also changes in capitalized/non-capitalized rental income, from the construction of price index of Jeonse, the dominant lease structure in the housing market in South Korea.
Accordingly, a machine learning model based on two of the most well-established algorithms, namely LightGBM (Light Gradient Boosting Machine) and ANN (Artificial Neural Network), is trained using complete transaction data of Seoul Apartments sales and Jeonse contracts from 2011 to 2024. This enables the estimation of both housing asset price and Jeonse price changes at the individual housing unit level. Unlike traditional surveyed or appraised-based repeat sales models and hedonic price models, this index approach allows for the construction of price indices with high spatial and temporal resolution. The key empirical contribution of this study lies in its ability to assess the inflation hedging performance of both capital and rental returns at the micro level of individual housing units, rather than at aggregated national or metropolitan scales. This, in turn, enables a more granular estimation of inflation risk.
As a result, the resulting house price and Jeonse price indices cover 2,160 apartment complexes in Seoul, encompassing 1,627,871 housing units, which account for approximately 92.4% of the total apartment stock in Seoul. The model performance for the House Price Automated Valuation Model (AVM) showed a Mean Absolute Percentage Error (MAPE) of 4.54% to 4.91%, while the Jeonse Price AVM recorded a MAPE of 7.62% to 7.83%.
Furthermore, the study examines potential discrepancies embedded in the official CPI figures and proposes adjusted variables of the Consumer Price Index that more accurately and promptly reflect housing cost component. These adjustments incorporate actual transaction-based rent trends and owner-occupied housing costs in place of survey-based estimates. The study compares several inflation series for Seoul, including the official CPI with housing costs, the CPI excluding housing costs, and two adjusted CPI measures that use actual transaction-based rent data. The results indicate that the official CPI increased by an average of 1.94% per year from 2011 to 2024, while the non-housing CPI rose by 1.95%, suggesting that housing cost component have a suppressing effect on the current overall CPI. When the surveyed-based rent index in the CPI was replaced with a 12-month rolling average of Seoul’s transaction-based rent index, the average annual increase rose to 2.17% per year. Moreover, when owner-occupied housing costs were also included, the adjusted CPI showed a further increase to 2.55% per year.
The first research hypothesis examines whether the inflation hedging effect of housing differs depending on the type of house price index used, such as those based on surveyed or appraised values, or actual transaction prices. Using a vector error correction model (VECM), the study finds that the conventional KB house price index, which relies on appraised prices, implies only a modest hedging effect of approximately 0.27 percent in response to a 1 percent rise in inflation. In contrast, the machine learning-based transaction price index indicates a much stronger hedge, with housing prices increasing by 2.60 percent approximately, thereby establishing a long-run cointegration relationship. These findings suggest that appraisal-based indices, which tend to smooth price movements, may significantly underestimate the inflation hedging capacity of housing assets and call for a re-examination of conclusions derived from appraisal-based data alone.
The second research hypothesis questions if the Jeonse prices, as proxy to capitalised rental income, also functions as inflation hedge. The second VECM focuses on rental returns by applying two types of Jeonse Price Indices (JPI): a standard transaction-based JPI and a cost-based JPI that represents opportunity cost of renting a house. These represent two approaches to capturing the rental return available to homeowners under Korea’s unique deposit-based lease system. The estimation results show that a 1% increase in inflation is associated with approximately 1.6195% increase in the JPI, and approximately 2.0368% increase in the cost-based JPI. These findings show that both the opportunity cost of Jeonse and the capital value of Jeonse deposits exhibit strong inflation-hedging properties. This suggests that landlords can pre-emptively set Jeonse prices to prevent future inflation risks, thereby reinforcing the role of Jeonse as a proactive means to inflation hedge.
Third hypothesis questions the degree to which housing asset hedges against inflation differs across localised submarkets within Seoul. The extent covers analysis annual real return across entire Seoul and major Seoul’s submarkets: the Gangnam and Gangbuk submarkets. While Seoul’s housing market showed a positive annual real return averaging 5.46% during 2012-2024 demonstrating consistent excess returns even after accounting for inflation, Gangnam submarkets outperformed with a 6.70% average, consistently delivering higher returns compared to Seoul and the Gangbuk submarket (4.71%). Considering the long-term holding behaviour of homeowners, the analysis by holding period of 3, 5, 7 and 9-years shows that the probability of achieving an annual capital return exceeding 5 percent increases as the holding period becomes longer. For Seoul as a whole, the probabilities are 57.52%, 67.41%, 83.61%, and 88.40% for the 3, 5, 7, and 9-year holding periods, respectively. In the Gangbuk submarket, the corresponding probabilities are 56.59%, 65.41%, 78.64%, and 86.86%, while in the Gangnam submarket, they are higher at 62.12%, 71.62%, 91.73%, and 93.45%. These findings suggest that the likelihood of achieving higher annual returns increases significantly with longer investment durations, with particularly pronounced probabilities observed in the Gangnam area.
Taken together, the findings indicate that both capital gains and rental income from housing assets in South Korea provide more than a one-to-one hedge against inflation. This underscores the effectiveness of real estate as an inflation protection tool, particularly when market-based measures are employed. It also highlights the importance of evaluating housing assets through both price dynamics and rent-setting behaviour when assessing their role as an inflation hedge.