South Korea Aggregate Distance-to-Default, 25th Percentile (Weak Tail)

The aggregate distance-to-default measures how many standard deviations of asset value separate the listed corporate sector from a default barrier, inferred from observable equity prices and disclosed balance-sheet debt. The construction follows the structural credit model of Merton (1974), which treats a firm's equity as a call option on its assets struck at its debt and so summarises credit risk in a single forward-looking statistic. The dashboard headline uses the naïve approximation of Bharath and Shumway (2008), nearly as predictive as the iterative-Merton solution while being robust and reproducible, with the default barrier set at short-term debt plus half of long-term debt per the KMV convention (Crosbie and Bohn 2003). The expected default frequency translates this distance into a one-year risk-neutral default probability, comparable across firms and over time:

A narrowing distance-to-default and a rising expected default frequency mean that corporate-sector default risk is rising.

KRED computes this structural model on Korean institutional inputs. The equity panel comes from the Korea Exchange for listed KOSPI and KOSDAQ nonfinancial firms, with financial firms excluded because their balance-sheet structure is inappropriate. The balance-sheet panel comes from Financial Supervisory Service disclosures, and because structured XBRL statements are reliably retrievable from 2015 onward while pre-2014 filings are not publicly retrievable, the sample begins in 2015 and covers the 2020 pandemic, the 2022 tightening cycle, and the late-2022 credit crunch triggered by the Legoland project-finance default without extending back to the 2008 global financial crisis. Each statement is applied point-in-time and is never used before its disclosure date, which removes look-ahead by construction, and delisted firms are included up to delisting so that the index is not survivorship-biased. The risk-free rate is the Bank of Korea 7-day base rate. The monthly cross-firm aggregation yields the median distance-to-default KRDD, the 25th-percentile distance-to-default KRDDTAIL, and the median expected default frequency KREDF.

KRDD is a directional early-warning signal for the listed-nonfinancial credit cycle rather than a substitute for bond-market spread data. That scope limitation is symmetric, as the late-2022 Legoland credit crunch was concentrated in non-listed project-finance and non-bank-financial-intermediary exposures that the listed-nonfinancial universe does not contain and so did not register in the index. The financials panel is current through the latest closed fiscal year on the annual cadence, while intra-year quarterly updates remain an ongoing task tied to the operational-pipeline expansion. The index is distinct from KRFCI, the broad financial-conditions index built on five market-stress variables, and from KRCGAP, the Basel III credit-cycle gauge, and it measures corporate-sector default risk inferred from equity valuations and disclosed filings rather than an excess bond premium that would require bond-level price data.

Sample updates are currently suspended, with the last observation stamped 30 June 2026, and later periods are not yet reflected. Whether and when updates resume is a separate data-pipeline decision.

Estimated using the eight-step Bharath and Shumway (2008) naïve approximation to the Merton (1974) structural model.

(1) Equity market value and volatility. For each listed nonfinancial firm $i$ and trading day $t$,

where $P_{i,t}$ is the close price and $\text{shares}_{i,t}$ is the shares outstanding per the KRX market-cap snapshot. The annualised equity return volatility is the 252-day rolling standard deviation of daily log returns,

(2) KMV default barrier. For each firm $i$ and most recent disclosed quarterly filing whose disclosure date $\leq t$,

where $\text{STD} =$ short-term borrowings $+$ current portion of long-term debt and $\text{LTD} =$ long-term borrowings, per Crosbie and Bohn (2003). If the borrowings lines are all absent, $D$ falls back to total liabilities (the robustness variant).

(3) Point-in-time discipline. The most recent filing at $t$ is the one whose disclosure date $\leq t$, where the disclosure date is the first eight digits of the disclosure receipt number. A statement is never used before its disclosure date; a unit test verifies that adding a future filing does not change any past month's $D$ (the hard look-ahead-free gate).

(4) Naïve asset volatility (Bharath and Shumway 2008).

(5) Distance-to-default with risk-neutral drift. Using $\mu = r_t$ (the BOK 7-day base rate) and horizon $T = 1$ year,

(6) Expected default frequency.

(7) Cross-firm aggregation. Each calendar-month-end the cross-firm panel yields

The headline panel is the full listed-nonfinancial universe (KOSPI + KOSDAQ), equal-weighted; a KOSPI-only panel and an asset-weighted variant (weights $w_i \propto E_i + D_i$) are stored alongside.

(8) Burn-in and low-confidence flag. Each month is flagged low-confidence when the contributing firm count falls below 30 or during the 252-day volatility burn-in. The Phase-1 seed panel of 45 well-known KOSPI+KOSDAQ blue chips delivers a Jan-2021 through Jun-2025 window with 41 to 44 firms per month at the headline (no low-confidence months on the headline panel).

The aggregate distance-to-default is the canonical equity-market-based read on the corporate-sector credit cycle in the Merton (1974) structural credit framework. Bharath and Shumway (2008) document in a panel of US firms that the naïve distance-to-default is highly predictive of subsequent realised defaults at the one-year horizon, the same horizon $T = 1$ that the dashboard uses, and that adding the iterative Merton solution gains little incremental information. Vassalou and Xing (2004) document that the cross-section of distance-to-default is priced in equity returns, with high-default-risk firms commanding a risk premium that the Black and Scholes (1973) option-pricing intuition embeds in the structural model. Campbell, Hilscher, and Szilagyi (2008) extend the analysis to reduced-form predictors and show that the Merton-based gauge enters the most-predictive specification as a load-bearing input, with Duffie, Saita, and Wang (2007) embedding it in a multi-period default-prediction framework that is now standard in central-bank financial-stability monitoring.

The cross-firm median distance-to-default condenses the corporate sector's credit risk into a single forward-looking statistic that responds quickly to equity-market moves, the property the International Monetary Fund Global Financial Stability Report exploits in its corporate-DD distribution monitoring, the European Central Bank Financial Stability Review uses for bank-DD distributions across the euro area, the Bank of England Financial Stability Report tracks for UK financials and nonfinancials, and the Bank of Korea Financial Stability Report adopts in its semiannual corporate-vulnerability assessments. Aikman, Haldane, and Nelson (2015) frame this distribution reading inside the macroprudential framework that Basel III countercyclical-capital-buffer activation operationalises, with the joint reading of the aggregate distance-to-default and KRCGAP providing the credit-stock side and the corporate-default-risk side of the same medium-frequency financial cycle.

The dashboard separates the median KRDD from the 25th-percentile KRDDTAIL because the median misses the vulnerable tail of firms whose distress drives systemic-risk dynamics. Acharya et al. (2017) frame the systemic-risk decomposition in terms of the joint distribution of firm-level distress, and the 25th-percentile reading is the dashboard's quick statistic for the same idea on the equity-market-based gauge. Adrian and Brunnermeier (2016) embed this distribution-of-distress reading inside the CoVaR framework that the IMF GFSR adopts for cross-country comparison. The Bank of Korea Financial Stability Report routinely tracks the share of corporates whose interest-coverage ratio drops below one as a complementary panel-distress indicator that maps to the KRDDTAIL signal.

KRDD is negatively correlated with rating-bucket corporate-bond spreads within a structurally predicted band, complementing rather than substituting for spread data. On the full-panel 2016 through 2026 sample the first-difference correlation between KRDD and the AA corporate-versus-KTB spread is approximately negative one-sixth, strengthening to approximately negative one-half at the long-run level that carries the economically dominant common-cycle component. This magnitude sits inside the structural band $|\Delta\rho| \approx 0.10$ to $0.25$ that the credit-risk literature predicts between an aggregate equity-market-based default measure and rating-bucket bond spreads (Bharath and Shumway 2008; Collin-Dufresne, Goldstein, and Martin 2001; Gilchrist and Zakrajšek 2012; Campbell, Hilscher, and Szilagyi 2008).

KRDD is not the Gilchrist and Zakrajšek (2012) excess bond premium, which requires bond-level prices, per-issue duration-matched Treasury benchmarks, and CDS spreads that are not freely available for Korean issuers, but an equity-market-based alternative faithfully built from the free data of KRX equity prices and disclosed financials. A falling KRDD reads as the equity market pricing in deteriorating corporate-sector solvency, while the spread widening that KRFCI's credit contribution captures from the Bank of Korea-published AA−/BBB− 3-year spreads reads as the bond market pricing in the same, so the two channels are complementary.

Financial firms are excluded at the panel-construction step because their balance sheets are dominated by financial-instrument liabilities and their default mechanisms involve regulator-driven going-concern interventions that the Merton structural setup does not capture (Merton 1977). The Moody's KMV practice described by Crosbie and Bohn (2003) likewise excludes banks and insurers as a standard convention, so the aggregate KRDD reading is best read as the corporate-sector-credit-risk channel, complementary to the bank-sector financial-conditions channel that KRFCI's basis and credit contributions capture from the wholesale-funding side. The Merton structural model is itself an idealisation that assumes lognormal asset values, a single one-year horizon, and a single debt barrier, and the bond-pricing and credit-default-swap literature reports that this mapping is empirically tight on the cross-section but loose on the level of credit spreads (Eom, Helwege, and Huang 2004).

On the 45-firm seed panel the median KRDD declined from a healthy 4.5σ in the late-2022 credit-stress window to 2.86σ in 2025 Q2, KRDDTAIL fell from 2.44σ to 1.36σ, and median KREDF rose from 0.04 percent to 0.21 percent annually. This trajectory is consistent with the Korean macro backdrop of weakening corporate-sector earnings into 2025 and is driven by the seed-sample composition (Samsung Electronics 005930, SK Hynix 000660, and other KOSPI heavyweights) that is sensitive to the global semiconductor cycle and external-demand dynamics. Because the closest macro-credit-risk comparison, the Gilchrist and Zakrajšek (2012) excess bond premium, requires bond-level prices that are not freely available for Korean issuers, the equity-market-based KRDD framework provides the most faithful free-data alternative.

KRDD is most useful read alongside KRCGAP, KRDSR, KRFCI, and KRKTBNOISE. KRCGAP captures the credit-stock dimension following Borio and Lowe (2002), KRDSR captures the debt-service cash-flow dimension following Drehmann and Juselius (2014), KRFCI captures the financial-conditions market-stress dimension, KRKTBNOISE captures the arbitrage-capital-limits dimension, and KRDD captures the corporate-sector default-risk dimension inferred from equity-market valuations. Together the five series provide a layered read on the financial cycle that no single series can deliver alone.

KRDD is the canonical equity-market-based gauge of Korean corporate-sector credit risk that fixed-income desks operating in Korean credit markets reference. Its level relative to its own history rather than any numerical threshold identifies the corporate-credit-cycle position, with Bharath and Shumway (2008) showing that the cross-firm distribution of distance-to-default predicts subsequent realised defaults at the one-year horizon and Vassalou and Xing (2004) documenting the risk premium embedded in the cross-section of equity returns. A falling KRDD signals a tightening of corporate-sector credit risk that, as the Gilchrist and Zakrajšek (2012) framework formulated on US data shows, transmits to wider corporate-bond credit spreads and to slower corporate-sector earnings growth that drags on the policy-rate cycle.

For investors with positions in Korean corporate bonds, particularly high-yield and lower-rated names, the 25th-percentile KRDDTAIL is the natural read on the most-vulnerable quartile of nonfinancials. The dashboard exposes this tail statistic alongside the median KRDD because the median misses the vulnerable tail of firms whose distress drives systemic-risk dynamics. Acharya et al. (2017) frame the systemic-risk decomposition in terms of the joint distribution of firm-level distress, and the 25th-percentile reading is the dashboard's quick statistic for the same idea on the equity-market-based gauge.

A falling KRDD that coincides with a widening AA−/BBB− corporate-versus-KTB spread reads as the equity market and the bond market pricing in the same corporate-credit-risk deterioration, the highest-conviction signal of a credit-cycle turn. For positions sensitive to Korean credit risk (corporate-bond portfolios, bank-debt holdings, and Korean-financial-sector equity), this joint signal is the leading indicator the position should pre-emptively hedge against. Conversely, a rising KRDD that coincides with a tightening AA−/BBB− spread reads as a cyclical low at which marginal allocation to corporate credit is justified.

KREDF is the risk-neutral one-year default probability under $\mu = r$, a measure directly comparable across firms and over time (Bharath and Shumway 2008). For investors building corporate-bond and credit-default-swap portfolios in Korea, KREDF is the natural input to expected-loss models and to portfolio-level credit-VaR computations. Duffie, Saita, and Wang (2007) embed the same Merton-based hazard rate in a multi-period default-prediction framework that is standard in the credit-risk practitioner literature, and Campbell, Hilscher, and Szilagyi (2008) confirm its predictive value alongside reduced-form covariates.

The aggregate distance-to-default leads bank-sector loan-loss provisioning at roughly the one-to-two-year horizon, a property the European Central Bank Financial Stability Review documents for euro-area banks and the Bank of England Financial Stability Report tracks for UK banks. For investors with positions in Korean bank debt and equity, KRDD is the leading indicator of the credit-loss cycle that drives bank-sector earnings and capital adequacy. The 2020 pandemic episode is the canonical example of this transmission, with the corporate-sector distance-to-default falling sharply in March 2020 before transmitting to bank-sector loan-loss recognition over the following six to nine quarters, as Korean banks reported in 2021 and 2022.

Macroprudential-policy timing is formulated as the joint reading of a credit-stock-position gauge and the asset-price-cycle position, and for Korea the KRDD aggregate distance-to-default is the equity-market-based companion to the credit-stock gauge that KRCGAP captures (Aikman, Haldane, and Nelson 2015). For investors anticipating Bank of Korea macroprudential policy, the joint KRDD-and-KRCGAP signal is the best leading indicator of countercyclical-capital-buffer activation and of borrower-level tightening cycles. The 2017 through 2022 macroprudential cycle, in which the Financial Services Commission progressively tightened LTV, DTI, and DSR borrower limits in response to a rising KRDSR alongside a falling KRDD, illustrates this cross-channel reading.

Evidence that distance-to-default leads credit risk has accumulated across countries, with Gropp, Vesala, and Vulpes (2006) documenting that it leads CDS-based credit risk at the bank-sector level on euro-area bank data and Hillegeist et al. (2004) confirming the same leading property on US data. These findings translate to Korea given the institutional similarities documented in the Bank for International Settlements Annual Economic Reports and the Bank of Korea Financial Stability Reports. The KRDD aggregate index therefore enters the Korean macroprudential and credit-cycle reading as the equity-market-based corporate-credit-risk gauge that the Drehmann, Borio, and Tsatsaronis (2012) financial-cycle framework specifies and that Merton (1974), Bharath and Shumway (2008), and Crosbie and Bohn (2003) jointly operationalise.

KRDD is most useful read alongside KRFCI and KRCP. KRFCI captures the broad financial-conditions dimension built on five market-stress variables, KRCP captures the Cochrane-Piazzesi bond-return-forecasting dimension, and KRDD captures the corporate-sector default-risk dimension inferred from equity-market valuations. Bharath and Shumway (2008), Duffie, Saita, and Wang (2007), and Campbell, Hilscher, and Szilagyi (2008) all emphasise that the equity-market-based distance-to-default complements rather than substitutes for bond-market-based credit-risk measures, which is the framing the dashboard adopts in combining the three series into a layered read that no single series can deliver alone.

This is a model-derived series, the output of the Merton structural distance-to-default aggregation, so the unit-root reading describes the fitted measure rather than a directly observed price, and the serial-correlation and break diagnostics run on the first difference. The series is tested over its own span from 2016-01-31 to 2026-06-30.

The tail distance to default gives an ambiguous reading on the level, since the Dickey and Fuller (1979) and Phillips and Perron (1988) tests reject a unit root at p = 0.1947 and p = 0.0 while the Kwiatkowski et al. (1992) test also rejects stationarity, so no clean order is assigned. On the first difference the Ljung and Box (1978) portmanteau rejects white noise at lags 12 and 24, Q = 251.90 and Q = 353.48 at p = 0.000 and p = 0.000. The Bai and Perron (1998, 2003) procedure finds no break in the mean, a reading consistent with the parameter-instability inference of Andrews (1993) on the differenced object (Perron 1989).

This is a monthly model output with no posited low-integer seasonal component, so the seasonal machinery of Hylleberg et al. (1990) and Canova and Hansen (1995) carries no meaningful object and is not run (Beaulieu and Miron 1992; Ghysels and Osborn 2001).