---
ticker: "KRRRGAP"
title: "South Korea Real Rate Gap"
unit: "%p"
frequency: "Quarterly"
source: "Holston, Laubach, and Williams (2023)"
release: "Updated Quarterly"
category: "Real & Policy Rates"
country: "KR"
language: "en"
canonical: "https://kred.dev/en/series/KRRRGAP"
license: "https://creativecommons.org/licenses/by-nc-nd/4.0/"
latest_value: 0.06
latest_date: "2026-06-30"
first_date: "1997-03-31"
observations_total: 118
observations_shown: 118
---

# South Korea Real Rate Gap

## Overview

The distance between the real policy rate and neutral, measuring whether conditions are easy or tight.

## Key Figures

|  | Value | Date |
|---|---|---|
| Latest | 0.06 | 2026-06-30 |
| Change from previous | -0.12 | 2026-03-31 |
| Change over one year | -0.49 | 2025-06-30 |
| Highest on record | 18.88 | 1998-03-31 |
| Lowest on record | -2.99 | 2009-03-31 |
| Period covered | 1997-03-31 – 2026-06-30 |  |
| Observations | 118 |  |

## Recent observations

| Date | Value | Change |
|---|---|---|
| 1997-03-31 | 4.25 |  |
| 1997-06-30 | 5.34 | +1.09 |
| 1997-09-30 | 5.89 | +0.55 |
| 1997-12-31 | 10.55 | +4.66 |
| 1998-03-31 | 18.88 | +8.33 |
| 1998-06-30 | 9.30 | -9.58 |
| 1998-09-30 | 1.36 | -7.94 |
| 1998-12-31 | -0.53 | -1.89 |
| 1999-03-31 | -0.98 | -0.44 |
| 1999-06-30 | 1.20 | +2.18 |
| 1999-09-30 | 1.99 | +0.79 |
| 1999-12-31 | 1.78 | -0.22 |
| 2000-03-31 | 1.58 | -0.20 |
| 2000-06-30 | 0.82 | -0.76 |
| 2000-09-30 | 0.39 | -0.43 |
| 2000-12-31 | -0.20 | -0.59 |
| 2001-03-31 | -0.74 | -0.55 |
| 2001-06-30 | -1.46 | -0.72 |
| 2001-09-30 | -2.44 | -0.98 |
| 2001-12-31 | -2.46 | -0.02 |
| 2002-03-31 | -2.13 | +0.33 |
| 2002-06-30 | -1.39 | +0.74 |
| 2002-09-30 | -1.41 | -0.02 |
| 2002-12-31 | -1.47 | -0.06 |
| 2003-03-31 | -1.79 | -0.33 |
| 2003-06-30 | -1.92 | -0.13 |
| 2003-09-30 | -2.05 | -0.13 |
| 2003-12-31 | -1.77 | +0.28 |
| 2004-03-31 | -1.28 | +0.49 |
| 2004-06-30 | -1.02 | +0.27 |
| 2004-09-30 | -0.73 | +0.29 |
| 2004-12-31 | -1.30 | -0.57 |
| 2005-03-31 | -1.13 | +0.16 |
| 2005-06-30 | -1.47 | -0.33 |
| 2005-09-30 | -1.20 | +0.27 |
| 2005-12-31 | -0.51 | +0.68 |
| 2006-03-31 | -0.24 | +0.27 |
| 2006-06-30 | 0.46 | +0.70 |
| 2006-09-30 | 0.51 | +0.05 |
| 2006-12-31 | 0.49 | -0.01 |
| 2007-03-31 | 0.46 | -0.03 |
| 2007-06-30 | 0.33 | -0.13 |
| 2007-09-30 | 0.69 | +0.36 |
| 2007-12-31 | 0.96 | +0.27 |
| 2008-03-31 | 1.03 | +0.07 |
| 2008-06-30 | 0.90 | -0.12 |
| 2008-09-30 | 0.72 | -0.19 |
| 2008-12-31 | -0.84 | -1.55 |
| 2009-03-31 | -2.99 | -2.15 |
| 2009-06-30 | -2.81 | +0.19 |
| 2009-09-30 | -2.08 | +0.72 |
| 2009-12-31 | -1.70 | +0.38 |
| 2010-03-31 | -1.37 | +0.33 |
| 2010-06-30 | -1.12 | +0.24 |
| 2010-09-30 | -0.55 | +0.58 |
| 2010-12-31 | -0.42 | +0.13 |
| 2011-03-31 | 0.17 | +0.59 |
| 2011-06-30 | 0.02 | -0.15 |
| 2011-09-30 | -0.26 | -0.28 |
| 2011-12-31 | -0.27 | -0.01 |
| 2012-03-31 | -0.19 | +0.08 |
| 2012-06-30 | 0.31 | +0.50 |
| 2012-09-30 | 0.82 | +0.52 |
| 2012-12-31 | 0.82 | 0.00 |
| 2013-03-31 | 0.69 | -0.13 |
| 2013-06-30 | 0.48 | -0.21 |
| 2013-09-30 | 0.34 | -0.14 |
| 2013-12-31 | 0.43 | +0.09 |
| 2014-03-31 | 0.29 | -0.14 |
| 2014-06-30 | 0.41 | +0.12 |
| 2014-09-30 | 0.01 | -0.39 |
| 2014-12-31 | -0.35 | -0.37 |
| 2015-03-31 | -0.02 | +0.34 |
| 2015-06-30 | -1.19 | -1.18 |
| 2015-09-30 | -1.22 | -0.03 |
| 2015-12-31 | -1.47 | -0.25 |
| 2016-03-31 | -1.53 | -0.06 |
| 2016-06-30 | -0.90 | +0.63 |
| 2016-09-30 | -1.29 | -0.39 |
| 2016-12-31 | -1.15 | +0.14 |
| 2017-03-31 | -0.99 | +0.16 |
| 2017-06-30 | -1.01 | -0.02 |
| 2017-09-30 | -0.82 | +0.19 |
| 2017-12-31 | -0.66 | +0.15 |
| 2018-03-31 | -0.49 | +0.17 |
| 2018-06-30 | -0.22 | +0.27 |
| 2018-09-30 | -0.42 | -0.20 |
| 2018-12-31 | -0.04 | +0.38 |
| 2019-03-31 | 0.06 | +0.10 |
| 2019-06-30 | 0.19 | +0.13 |
| 2019-09-30 | 0.29 | +0.10 |
| 2019-12-31 | -0.04 | -0.33 |
| 2020-03-31 | -0.05 | -0.01 |
| 2020-06-30 | -0.39 | -0.35 |
| 2020-09-30 | -0.39 | 0.00 |
| 2020-12-31 | -0.60 | -0.21 |
| 2021-03-31 | -0.46 | +0.14 |
| 2021-06-30 | -0.89 | -0.44 |
| 2021-09-30 | -1.12 | -0.23 |
| 2021-12-31 | -0.97 | +0.16 |
| 2022-03-31 | -1.34 | -0.38 |
| 2022-06-30 | -1.53 | -0.19 |
| 2022-09-30 | -1.41 | +0.12 |
| 2022-12-31 | -1.21 | +0.20 |
| 2023-03-31 | -0.87 | +0.34 |
| 2023-06-30 | -0.63 | +0.24 |
| 2023-09-30 | -0.35 | +0.28 |
| 2023-12-31 | 0.16 | +0.51 |
| 2024-03-31 | 0.42 | +0.26 |
| 2024-06-30 | 0.82 | +0.41 |
| 2024-09-30 | 1.08 | +0.25 |
| 2024-12-31 | 0.92 | -0.16 |
| 2025-03-31 | 0.84 | -0.08 |
| 2025-06-30 | 0.55 | -0.29 |
| 2025-09-30 | 0.23 | -0.31 |
| 2025-12-31 | 0.47 | +0.24 |
| 2026-03-31 | 0.18 | -0.29 |
| 2026-06-30 | 0.06 | -0.12 |

## Definition

The real rate gap is the difference between the ex-ante real short-term interest rate and the natural rate of interest, measuring the stance of monetary policy relative to the economy's structural equilibrium.

Formally, the real rate gap is defined as:

$$\text{Real Rate Gap}_t = r_t - r^*_t$$

where

$$r_t = i_t - \pi^e_t$$

is the real interest rate, the MSB 91-day yield minus expected inflation, which within the HLW model is constructed as a 4-quarter trailing average of core CPI inflation, and

$$r^*_t = 4c\,g_t + z_t$$

is the natural rate.

The real rate gap directly enters the IS curve of the HLW model as the mechanism through which monetary policy affects the real economy:

$$\tilde{y}_t = a_{y,1}\,\tilde{y}_{t-1} + a_{y,2}\,\tilde{y}_{t-2} + \dfrac{a_r}{2}\left[(r_{t-1} - r^*_{t-1}) + (r_{t-2} - r^*_{t-2})\right] + \epsilon^{\tilde{y}}_t \tag{IS}$$

The coefficient $a_r < 0$ ensures that a positive real rate gap reduces the output gap, consistent with the standard monetary transmission mechanism.

A positive gap ($r_t > r^*_t$) indicates that monetary policy is tighter than neutral, where the real interest rate exceeds the rate consistent with balanced growth, restraining aggregate demand and putting downward pressure on inflation. A negative gap ($r_t < r^*_t$) indicates accommodative policy.

## Methodology

Estimated jointly with the natural rate of interest ($r^*$, KRNR), trend growth (KRTRGDP), and the output gap (KRGDPGAP) within the HLW (2023) 3-stage MLE state-space model. The full model specification, data inputs, and estimation procedure are described in detail in the KRNR methodology section.

**(1) Real rate gap computation.** The real rate gap is not a Kalman filter state variable but is computed post-estimation as the difference between the observed real rate and the smoothed natural rate:

$$\text{Real Rate Gap}_t = r_t - r^*_t = (i_t - \pi^e_t) - (4c\,g_t + z_t) \tag{RRG}$$

where $i_t$ is the MSB 91-day yield (quarterly average of daily observations), $\pi^e_t$ is the 4-quarter trailing average of core CPI inflation (serving as the expected inflation proxy within the HLW framework), $g_t$ is the Kalman-smoothed trend growth, and $z_t$ is the smoothed residual natural rate component. All components except the nominal interest rate and observed inflation are model outputs.

**(2) Expected inflation specification.** The HLW model uses backward-looking inflation expectations by design:

$$\pi^e_t = \dfrac{1}{4}\sum_{j=1}^{4}\pi_{t-j}$$

that is, a 4-quarter trailing average of past core CPI inflation rates. This is consistent with the original Laubach and Williams (2003) and HLW (2017, 2023) specification, where the IS and Phillips curve equations are jointly estimated under the assumption that agents form expectations by extrapolating recent inflation experience. This differs from KRED's standalone ex-ante real rate series KREARPR, which uses a forward-looking expected inflation estimate combining UCSV trend inflation (Stock and Watson 2007) with BOK Consumer Survey data via Nelson-Siegel. The backward-looking specification in HLW ensures internal consistency within the structural model, where the same $\pi^e_t$ enters both the real rate construction and the Phillips curve, so that the IS-curve parameter $a_r$ is estimated conditional on this specific expectations process.

**(3) Estimation uncertainty.** The real rate gap combines observable and unobservable components. The nominal rate $i_t$ and inflation $\pi_t$ are observed, but $r^*_t$ is estimated with substantial uncertainty from the Kalman smoother. The gap therefore inherits the full estimation uncertainty of the natural rate:

$$se(\text{gap}_t) = se(r^*_t) = \sqrt{c^2 \cdot se(g_t)^2 + se(z_t)^2}$$

A real rate gap of, say, +0.5% has limited informational content if the standard error of $r^*$ is ±1.5%. The sign and magnitude of the gap should be interpreted in light of this uncertainty, particularly at the end of the sample where smoother estimates are least reliable.

## Applications in Economics

The real rate gap is the most direct measure of the monetary policy stance available from the HLW framework, answering the question of whether monetary policy is stimulative or restrictive relative to the economy's equilibrium.

The real rate gap is closely related to the Taylor (1993) rule framework. In a standard Taylor rule, the policy rate responds to the inflation gap and the output gap

$$i_t = r^* + \pi^* + \alpha(\pi_t - \pi^*) + \beta \tilde{y}_t$$

Rearranging, the implied real rate gap is

$$r_t - r^* = (\alpha - 1)(\pi_t - \pi^*) + \beta \tilde{y}_t$$

The HLW real rate gap provides a model-based estimate of this deviation, jointly estimated with the output gap and the natural rate, allowing researchers to assess whether the BOK's policy actions have been broadly consistent with a Taylor-type rule or whether systematic deviations have occurred.

The real rate gap provides a narrative lens for Korean monetary policy history. Following the 1997–98 Asian financial crisis, interest rates were sharply elevated (positive real rate gap), contributing to the recession but also stabilizing the currency. The 2001–2005 period saw accommodative policy (negative gap) supporting the recovery. The 2008 global financial crisis led to aggressive easing (large negative gap), followed by normalization. During 2020, the BOK cut rates to 0.50% in response to COVID-19, and the real rate gap indicates whether this was sufficient to achieve an accommodative stance given the simultaneously declining $r^*$. The 2022–2023 tightening cycle brought the real rate gap sharply positive as the BOK responded to inflation.

Prolonged negative real rate gaps can fuel asset price inflation and financial imbalances, since they imply that monetary policy is persistently accommodative relative to the structural equilibrium. Korea's experience in the 2020–2021 period illustrates this risk, as deeply negative real rates coincided with rapid increases in housing prices, household borrowing, and speculative activity in equity and cryptocurrency markets. The real rate gap provides a structural framework for assessing this channel, where the duration and magnitude of a negative gap quantifies the degree of monetary accommodation that may feed financial stability risks. This connects to the broader debate on 'leaning against the wind' in monetary policy (Svensson 2017; Adrian and Liang 2018).

Unlike simple comparisons of the nominal policy rate to headline inflation, the real rate gap accounts for the time-varying equilibrium real rate. A 1% real rate may be deeply accommodative when $r^*$ is 2%, but restrictive when $r^*$ has fallen to 0%. This distinction is critical for Korea, where the natural rate has declined substantially over the past two decades. KRRRGAP measures the policy stance within the HLW model's internal logic, while KREARPR provides the model-free, forward-looking real rate more directly relevant for policy evaluation, using UCSV-based expected inflation rather than the HLW's backward-looking moving average. The two measures can diverge significantly during periods of rapid inflation regime change, when the backward-looking moving average lags the forward-looking UCSV estimate. For a model-free, forward-looking real rate, see KREARPR.

## Applications in Financial Markets

The real rate gap is a powerful indicator for fixed income strategy, FX positioning, and cross-asset allocation.

The real rate gap is a leading indicator of future monetary policy actions. Large positive gaps (tight policy) indicate the BOK has room to cut rates, as the current policy stance is more restrictive than neutral. This is bullish for short-duration KTB positions and IRS receivers. Negative gaps suggest eventual tightening, supporting short KTB positions and IRS payers. The gap's predictive power for BOK actions can be enhanced by combining it with the implied rate path from the KRED implied rate series, which captures market pricing of the next three Monetary Policy Committee meetings.

The real rate gap primarily affects the short end of the yield curve, as it measures the distance between the current policy rate and its equilibrium. When the gap is wide and positive, the front end is expected to rally as the BOK eases, favoring curve steepeners (long front-end, short back-end). When the gap is negative, the front end is expected to sell off as the BOK normalizes, favoring curve flatteners. The signal is most useful when combined with the output gap (KRGDPGAP) and the term premium (KRTP series) to distinguish between front-end-driven and back-end-driven curve moves.

The real rate gap differential between Korea and other economies is a structural determinant of KRW dynamics. A positive Korean real rate gap relative to the U.S. (Korean policy is tighter than equilibrium relative to U.S. policy stance) tends to support KRW through capital inflows attracted by above-equilibrium yields. Conversely, a negative differential makes KRW carry less attractive. For carry trade strategies, the real rate gap is more informative than the raw interest rate differential because it adjusts for the structural equilibrium. A high Korean rate that is below $r^*$ (negative gap) may not be sustainable, while a moderate rate that is above $r^*$ (positive gap) represents a durable carry advantage.

In Korea, the real rate gap is closely correlated with the housing price cycle. Persistently negative gaps have historically coincided with housing market overheating, particularly in the Seoul metropolitan area, as accommodative monetary conditions drive leveraged real estate investment. Fixed income investors can use the real rate gap as an input for assessing macro-prudential policy risk. When the gap has been deeply negative for an extended period and housing prices are rising, the probability of macro-prudential tightening (LTV/DTI restrictions, targeted taxes) increases, which can independently affect the interest rate environment and credit conditions.

## Statistical Tests

KRRRGAP is the real-rate gap of the HLW natural-rate model, a two-sided Kalman-filtered and RTS-smoothed latent state estimated over 117 quarters from 1997-03-31 to 2026-03-31, so the persistence figures below are stated in quarters. Because it is a two-sided smoother output, its measured serial correlation is dominated by the filter's gain rather than by any data-generating process.

For that reason the integration-order battery is deliberately not run. The augmented unit-root regression of Dickey and Fuller (1979) with the lag augmentation of Said and Dickey (1984), the semiparametric Phillips and Perron (1988) test, the KPSS stationarity test (Kwiatkowski et al. 1992), the efficient GLS-detrended test of Elliott, Rothenberg, and Stock (1996), and the modified M-tests of Ng and Perron (2001) are all excluded, because a two-sided smoother is a symmetric moving average whose gain, not the underlying process, drives the serial correlation an integration test reads, and the random-walk-versus-constant character of the latent state is not point-identified by the likelihood (Stock and Watson 1998; Orphanides and van Norden 2002). The level mean-break search of Bai and Perron (1998) is likewise not run, since its asymptotics require a stationary object and a filter-persistent path spuriously segments (Perron 1989).

The matrix-mandated replacement is a descriptive persistence summary, labeled as a property of the smoothed series rather than of the data. The lag-one autocorrelation and implied half-life are 0.807 and 3.24 quarters, so the real-rate gap decays far faster than the near-random-walk natural-rate level it is built from. These figures describe how slowly the smoothed path decays and carry no integration-order claim, and the strong lag-one autocorrelation a portmanteau such as that of Ljung and Box (1978) would register on this level is itself a property of the smoother's gain.

No order of integration is assigned, by ruling rather than by an inconclusive test. The lag-one coefficient carries the familiar downward small-sample bias near unity on a quarterly sample of 117 observations, so the half-life is a lower-leaning descriptive figure whose median-unbiased interval would follow the grid of Andrews (1993). Because the HLW loader rewrites the full state history on every re-estimation, the smoothed natural rate at any fixed past quarter revises across vintages and the real-rate gap inherits that revision, and no stored vintage panel exists to quantify the magnitude (Orphanides and van Norden 2002). The model-implied content is honest as an assumption rather than a discovery, namely the random-walk state equation for the natural rate whose smoothed path the gap is measured against and the median-unbiased signal-to-noise ratios the model pins by the method of Stock and Watson (1998).

## Frequently Asked Questions

### What is the Taylor-rule prescribed policy rate?

The output of a monetary policy reaction function mapping the natural rate, current inflation, the inflation target and the output gap into a nominal policy rate. It is a benchmark for discussion, not a recommendation.

### How is the natural rate of interest estimated?

The real rate consistent with output at potential and stable inflation is treated as an unobserved state and estimated in a state space model. It moves slowly and is revised backwards as new data arrive, so its most recent values carry the widest uncertainty.

### What does the real rate gap say about the policy stance?

The difference between the prevailing real policy rate and the estimated natural rate. A negative gap is conventionally read as an accommodative stance, subject to the wide confidence bands that accompany any natural rate estimate.