South Korea Jeonse Price Index

This group records Korea housing-market indicators as monthly not-seasonally-adjusted levels, each raw series recorded without transformation. The group holds two price indices, one for dwelling sale prices and one for jeonse leasehold-deposit prices, together with two quantity series, the stock of unsold new dwellings and the count of sale transactions.

The two price indices are fixed-basket aggregates, and because the reference weights are held across the comparison, movements in each index reflect price change alone rather than shifts in the stock composition (Laspeyres 1871). The current-weighted dual and the geometric-mean ideal index that satisfies the time- and factor-reversal tests bound that fixed-base level, and the expenditure-share log-change form is the superlative approximation against which it is read (Paasche 1874; Fisher 1922; Törnqvist 1936).

The concept each price index ultimately approximates is a constant-utility price ratio, made precise within the four-approach taxonomy of index-number theory and grounded in the exact and superlative aggregators (Konüs 1939; Diewert 1976, 2002). Because a housing basket is heterogeneous and slow to turn over, the recorded level departs from that target by a quality-change and composition wedge, and the axiomatic consistency the construction does satisfy belongs to the test-approach lineage (Diewert 1998; Boskin et al. 1996; Balk 1995).

The two quantity series are counts rather than prices, and they belong to the tradition that measures the housing cycle through turnover and inventory rather than valuation (Burns and Mitchell 1946; Mitchell and Burns 1938). Housing turnover and the overhang of unsold stock are classical coincident and lagging markers of that cycle (Persons 1923; Moore 1961). Because dwellings are the dominant collateral of household borrowing, these quantities also trace the leverage leg of the financial cycle (Schularick and Taylor 2012; Jordà, Schularick, and Taylor 2017; Borio 1997).

Each series is preserved as its raw distribution rather than a supply-shock-robust or smoothed cyclical central tendency (Eckstein 1981; Zarnowitz 1992).

KRED applies no transformation to these housing-market series, loading each published level as recorded with only a date parse and a reshape into the long store. It performs no re-indexing, deflation, seasonal adjustment, smoothing, annualizing, or year-over-year differencing, so a price index enters as its published index number and a quantity series enters as its published count.

The measurement basis by which the two price levels come to exist is fixed-basket aggregation, where a reference-period survey fixes the weights and the elementary price relatives are combined into the all-items index (Laspeyres 1871). That construction sits within the index-number framework whose alternatives are the current-weighted index, the symmetric ideal index, and the expenditure-share log-change index, the last two exact for flexible aggregators (Paasche 1874; Fisher 1922; Törnqvist 1936; Diewert 1976).

The four conceptual routes to the same price number are the fixed-basket, test, stochastic, and economic approaches, and the constant-utility concept the level targets is the ratio that holds housing service fixed (Diewert 2002; Konüs 1939). The gap between the recorded level and that target consists of quality-change and composition bias, and because the series is served exactly as measured KRED does not correct it, nor does it impose the axiomatic repairs of the test approach (Diewert 1998; Boskin et al. 1996; Balk 1995).

The two quantity series are formed by counting units rather than aggregating prices, the tabulation of turnover and inventory on which cyclical measurement of the housing market rests (Burns and Mitchell 1946; Mitchell and Burns 1938; Persons 1923). KRED records these counts as reported and applies neither the reference-cycle dating nor the trend extraction that would convert them into a smoothed cyclical indicator (Moore 1961; Zarnowitz 1992; Eckstein 1981).

As gauges of the housing market, these series carry the two faces of the sector that macroeconomic analysis reads together, price and quantity. The sale-price and jeonse indices measure the valuation of housing service, and their use as a stability gauge turns on how closely a fixed-basket level tracks a constant-utility price (Konüs 1939; Laspeyres 1871).

Because a housing basket is heterogeneous and turns over slowly, the recorded price level is a known biased proxy for the quality-constant concept, the composition and quality-change wedge that index-number theory attributes to a fixed-base construction (Diewert 1998; Boskin et al. 1996). Read against that wedge, the sale-price index still identifies the persistent valuation swing that policy responds to rather than a smoothed core (Eckstein 1981; Diewert 2002).

The unsold-stock and transaction series supply the quantity signal that leads and confirms those valuation swings, the turnover and inventory markers of the classical cycle (Burns and Mitchell 1946; Mitchell and Burns 1938). A rising overhang of unsold dwellings and a thinning of sale transactions are the coincident and lagging indicators through which a housing downturn is dated (Persons 1923; Moore 1961; Zarnowitz 1992).

For financial-stability analysis the four series are read as the housing leg of the credit cycle, since dwellings are the dominant collateral of household leverage and the amplitude of house-price and turnover cycles maps onto the build-up of that leverage (Schularick and Taylor 2012; Jordà, Schularick, and Taylor 2017; Borio 1997). The relative price of housing against the consumption basket frames the same signal in real terms (Fisher 1930).

For asset pricing the housing-market level is the observable to which mortgage credit, property-linked securities, and household balance sheets are marked. The two price indices settle the valuation leg, and index-number theory formalizes the deflation of nominal housing values through the time- and factor-reversal tests a price index should satisfy (Fisher 1922; Laspeyres 1871; Paasche 1874).

Because the level the market marks to is a fixed-basket construction rather than a constant-utility price, the composition and quality-change wedge is priced into the basis between a headline index and the repeat-sale value of an individual dwelling (Diewert 1998; Boskin et al. 1996; Törnqvist 1936). Whether successive vintages of the index chain coherently for settlement is governed by the axiomatic consistency of the construction (Balk 1995; Diewert 2002).

The unsold-stock and transaction series price the liquidity and supply-overhang risk that valuation alone omits, the turnover and inventory information on which the cyclical read of collateral rests (Burns and Mitchell 1946; Mitchell and Burns 1938; Moore 1961). A thinning transaction count is the market-liquidity warning that widens haircuts on property collateral well before the price index turns (Persons 1923; Zarnowitz 1992).

For macro positioning the four series are read as the housing leg of the financial cycle, whose amplitude and leverage build-up carry the tail risk that dwellings, as the dominant household collateral, transmit to the banking system (Schularick and Taylor 2012; Jordà, Schularick, and Taylor 2017; Borio 1997). The real housing value against which long-horizon returns are judged is the nominal level deflated by the consumption basket (Fisher 1930; Konüs 1939).

KRJEONSE is published as a not-seasonally-adjusted jeonse price index level, and the appropriate statistical object is its logarithm, which is expected to be integrated of order one. The sample is 272 monthly observations spanning 2003-11 to 2026-06, and the unit-root triplet is fitted on the log level with a constant and trend.

The augmented Dickey-Fuller regression of Dickey and Fuller (1979), in the lag-augmented form of Said and Dickey (1984), does not reject the unit root at p = 0.932, the test of Phillips and Perron (1988) likewise does not reject at p = 0.916, and the stationarity-null test of Kwiatkowski et al. (1992) rejects trend stationarity at p < 0.01, so all three legs agree on a clean I(1) reading rather than the spurious-stationarity artefact that would attend a near-unit moving-average error (Ng and Perron 2001).

Because the log level is an integrated object, the portmanteau and the break search are run on its first difference, since a level portmanteau on an integrated index reads mechanical near-unit autocorrelation and a level mean-break search spuriously segments a stochastic trend (Perron 1989; Hamilton 2018). The portmanteau statistic of Ljung and Box (1978), refining the form of Box and Pierce (1970), rejects the white-noise null on the differenced log level at p = 0.000, and the multiple-break procedure of Bai and Perron (1998), computed by the dynamic-programming algorithm of Bai and Perron (2003), locates four breaks in the mean of the differenced series at 2009-07, 2015-12, 2019-05, and 2022-09.

Seasonal tests are appropriate on this unadjusted monthly index. The seasonal-unit-root test of Hylleberg et al. (1990), with the monthly mechanics of Beaulieu and Miron (1993), rejects unit roots at the seasonal frequencies at p = 0.000, while the joint seasonal-dummy F does not reject at p = 0.356, so no stochastic seasonality survives and no jointly significant deterministic seasonal pattern is present, the muted calendar seasonality expected of a leasehold-deposit price level rather than the pronounced seasonality of the stationary-seasonality frame (Canova and Hansen 1995) and consistent with the not-seasonally-adjusted character of the series (Ghysels and Osborn 2001).