The Impact of Financial Stability Policies on Korea Treasury Bond Yields

A. Securities Firms’ Call Market Restrictions

Table 1 presents changes in key institutional arrangements related to Korea’s major short-term financial markets.

TABLE 1

MAJOR INSTITUTIONAL CHANGES IN THE SHORT-TERM FINANCIAL MARKET IN KOREA

Restrictions on securities firms' participation in the call market progressed as follows. Initially, on July 27, 2010, the government announced that call borrowing by securities firms would be limited to 100% of their equity capital, implementing this regulation from October 1, 2010. Figure 1 demonstrates that securities firms' call borrowing ratios relative to equity capital (hereafter referred to as the call borrowing ratio) were already below 100% before the regulation. However, following the announcement (①) and implementation (②), the average and median ratios steadily declined. Numbers ① to ⑥ represent the sequence of announcements or implementations of regulations on securities firms’ call market participation. Additionally, although the maximum value of call borrowing ratios temporarily rose in the quarter immediately following implementation, it subsequently exhibited a downward trend.

FIGURE 1.

SECURITIES FIRMS’ CALL BORROWING RATIOS RELATIVE TO THEIR EQUITY CAPITAL

Note: ① Announcement of reduction in securities firms’ call borrowing limit to within 100% of their equity capital (July 27, 2010); ② Implementation of the 100% limit (October 1, 2010); ③ Announcement of further reduction in call borrowing limit to within 25% of equity capital (June 22, 2011); ④ Implementation of the 25% limit (July 12, 2012); ⑤ Announcement of complete exclusion of secondary financial institutions from the call market (November 22, 2013); ⑥ Implementation of the exclusion (February 24, 2015).

Source: Financial Supervisory Service, Financial Statistics Information System.

Subsequently, the government announced on June 22, 2011 (③) that it would reduce the securities firms’ call borrowing limit further to 25% of equity capital, implementing this measure on July 12, 2012 (④). Before the reduction announcement in the second quarter of 2011, the average and median borrowing ratios slightly increased, but from the third quarter of 2011 onward, a consistent decline was observed. The distribution of securities firms’ call borrowing ratios decreased sharply. The maximum call borrowing ratio among securities firms was 93.6% in the first quarter of 2010 but dropped to 40.5% by the third quarter of 2012 when the 25% regulation was implemented.

Finally, on November 22, 2013 (⑤), the government declared the exclusion of secondary financial institutions from the call market, implementing this rule from February 24, 2015 (⑥). After the regulatory announcement in the fourth quarter of 2013, declines in both the average and median call borrowing ratios accelerated significantly compared to the period before the announcement. Although the distribution of ratios did not change substantially immediately following the announcement (⑤), it fell dramatically after implementation (⑥). At the time of the announcement in the fourth quarter of 2013, the highest call borrowing ratio among securities firms was 46.4%, but this dropped sharply to 18.6% by the second quarter of 2015, following implementation.

Since the prohibition of call transactions for non-bank financial institutions in March of 2015, the volume of domestic call transactions has continued to decline, stabilizing at an average daily level of approximately ten trillion KRW. Meanwhile, the domestic Repurchase Agreement (RP) market has consistently expanded following government reforms in the call market, with the average daily outstanding volume reaching 147.53 trillion KRW as of August of 2022, an increase of 12.73 trillion KRW compared to the same month in the previous year. Figure 2 illustrates the average daily outstanding volume of inter-institutional RP purchases. Examining the short-term series reveals that RP transactions have steadily increased following the Bank of Korea’s market stabilization measures, including RP purchases initiated in April of 2020 to address financial market stability and liquidity concerns amid COVID-19.

FIGURE 2.

AVERAGE DAILY OUTSTANDING AMOUNT OF INTER-INSTITUTIONAL REPO PURCHASES

Note: 1) The dashed lines in the long-term time series indicate key regulatory events related to RP transactions: ① Prohibition of call transactions by non-bank financial institutions (Mar. 2015), ② Phase 1 of mandatory liquidity asset holdings (Jul. 1–31, 2020), ③ Phase 2 (Aug. 1, 2020 – Apr. 30, 2021), ④ Phase 3 (since May 1, 2021 – present); 2) The dashed lines in the short-term time series correspond to the same regulatory phases as ②–④ in Note 1.

Source: Korea Securities Depository, Yonhap Infomax.

Observing the steady increase in RP transactions depicted in Figure 2, we analyzed the phenomenon in which securities firms expanded their participation in the Repurchase Agreement (RP) market due to restrictions on their participation in the call market, as detailed in the analysis below. This development represents the initial phase of increased government bond demand and a subsequent reduction in government bond yields resulting from policies designed to enhance short-term money market stability. With their participation in the call market restricted, securities firms sought alternative funding channels, prompting greater involvement in the RP market. Unlike the call market, which facilitates borrowing without collateral, the RP market requires collateral, predominantly government bonds, which account for approximately 65% of the collateral utilized.

We provided empirical evidence through a difference-in-differences analysis to demonstrate the increase in RP market participation by securities firms due to restrictions on their participation in the call market. The call market participation restriction policy was implemented in three phases: initially limiting participation to no more than 100% of equity capital (October 1, 2010), subsequently restricting it to no more than 25% of equity capital (July 12, 2012), and finally excluding participation in the call market entirely (February 24, 2015). Because there was no available data for the first phase (October 1, 2010 – December 2011), we focused on analyzing the period during which participation was restricted to 25% of equity capital.

In the difference-in-differences regression analysis, securities firms directly affected by the policy formed the treated group, while asset management firms unaffected by the policy served as the control group. For unbiased estimates from this method, the parallel trends assumption between the treated and control groups needed to be satisfied. Figure 3 illustrates trends in the average daily RP sales (one-month moving average) by securities firms and asset management firms before and after the implementation of the 25% equity capital restriction policy. As depicted, prior to the policy, the average daily RP sales volumes of both groups moved parallel with a consistent gap. However, following the policy implementation, there was a significant increase in the RP sales volume by securities firms.

FIGURE 3.

RP PURCHASE TRENDS OF SECURITIES FIRMS AND ASSET MANAGEMENT COMPANIES

Note: The blue line represents the moving average (k = 61) of RP purchases by securities firms, and the red line represents the moving average (k = 61) of RP purchases by asset management companies. The dotted vertical line indicates the regulatory implementation date considered in the analysis, July 12, 2012.

Source: Bond Information Center, Korea Financial Investment Association.

A similar pattern was observed in relation to the complete exclusion of call market participation. Approximately two years after the implementation of the policy restricting call market participation to no more than 25% of equity capital, the level difference in call market participation between the treated and control groups expanded further. However, a situation arose again where the parallel trends assumption was satisfied. Subsequently, the complete exclusion policy from the call market was implemented, leading once more to a significant increase in the RP market participation gap between the treated and control groups.

Based on these observations, we employed a difference-in-differences regression analysis to estimate the extent of increased RP market participation resulting from these policies. In addition to the restrictions on securities firms’ call market participation, our analysis also incorporated various macroeconomic and financial-market-related variables that could influence funding demand through RP sales by securities firms and asset management companies. Among these, economic growth rates, industrial production indices, and inflation rates reflect macroeconomic conditions, while previous-day yields of the 3-year government bonds and CDS premiums represent financial market conditions. To avoid reverse-causality issues stemming from the correlation between RP sales volumes and demand for 3-year government bonds, we utilized the previous day’s yield. The difference-in-differences regression analysis model is formulated as follows:

The analysis results indicated that restricting securities firms' participation in the call market to no more than 25% of their equity capital led to an increase in their average daily RP sales volume by approximately 2.1 trillion KRW (Table 2). An increase of 2.1 trillion KRW in daily RP sales implies a corresponding rise in the demand for securities used as collateral, primarily government bonds, by approximately 1.365 trillion KRW (2.1 trillion KRW * 0.65). Given that the yield of 2-year government bonds reflects expectations over the following two years, it is also essential to consider extending the analysis period to two years following policy implementation. Table 3 presents the analysis results over this extended two-year period. According to these findings, the daily average RP sales volume of securities firms increased by approximately 5.0 trillion KRW over the two years after the implementation of the 25% equity capital restriction. Consequently, the increased demand for securities, including government bonds, due to the call market participation restriction is estimated to be at least 3.25 trillion KRW (5 trillion KRW * 0.65).

TABLE 2

DIFFERENCE-IN-DIFFERENCE REGRESSION RESULTS OF RP PURCHASES BY SECTOR OVER 180 BUSINESS DAYS FOLLOWING THE REGULATION IMPLEMENTATION

Note: 1) Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1; 2) The analysis period spans the dates of January of 2012 to April 2, 2013, covering 180 business days following the implementation of the regulation.

TABLE 3

DIFFERENCE-IN-DIFFERENCE REGRESSION RESULTS OF RP PURCHASES BY SECTOR OVER 720 BUSINESS DAYS FOLLOWING THE REGULATION IMPLEMENTATION

Note: 1) Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1; 2) The analysis period spans the dates of January of 2012 to June 11, 2015, covering 720 business days following the implementation of the regulation.

B. Bank LCR Requirement

Following the Global Financial Crisis of 2008-09, the Basel Committee on Banking Supervision (BCBS) introduced regulations to strengthen banks' liquidity management. These regulations consist of the liquidity coverage ratio (LCR), which addresses short-term liquidity management, and the net stable funding ratio (NSFR), which targets medium- to long-term liquidity stability.

Specifically, the LCR regulation requires banks to maintain a certain proportion of high-quality liquid assets to cover net cash outflows over a 30-day stress period. The definition of the LCR regulation is as follows:

The introduction of policies related to the liquidity coverage ratio (LCR) proceeded as described below. In December of 2010, the Basel Committee on Banking Supervision (BCBS) introduced a draft of Basel III, adding two liquidity regulations—the liquidity coverage ratio (LCR) and the net stable funding ratio (NSFR)—to the existing Basel II framework, which focused primarily on bank capital ratio regulations. Initially, the draft mandated banks to maintain an LCR of at least 100% starting from January of 2015. However, in January of 2013, BCBS relaxed this requirement, setting the LCR at 60% in January of 2015 with a planned annual increment of 10%, reaching full compliance by 2019.

Following this, major economies announced their implementation plans for Basel III, and the Korea government also adopted the LCR regulations in stages. In August of 2014, Korean financial authorities announced amendments to banking supervision regulations, stipulating that, effective January 1, 2015, domestic banks were required to maintain an LCR of at least 100%. This standard was stricter than the revised BCBS guidelines, reflecting the fact that many banks in Korea were already maintaining an LCR above 100%. This stricter standard likely stemmed from Korea's existing regulations on the Korean won liquidity ratio (liquid assets/liquid liabilities with maturities of one month or less), similar to the LCR (Bank of Korea, 2014). However, in December of 2014, the authorities relaxed the LCR requirement to 80%, announcing incremental annual increases of five percentage points until reaching full compliance by 2019.

Subsequently, the LCR requirement remained at 100% until April of 2019, when the Korean government relaxed the LCR regulations in response to the COVID-19 crisis. Specifically, the financial authorities temporarily lowered the minimum LCR requirement from 100% to 85% for six months, enabling banks to utilize their high-quality liquid assets as a means of crisis response (Financial Services Commission and Financial Supervisory Service, “Measures for Flexibility in Financial Regulations in Response to COVID-19,” announced April 16, 2020). This temporary easing of the LCR regulations, initially set to expire in September of 2020, was extended three times, each for six months, lasting until March of 2022.

From June of 2022, authorities initiated a gradual normalization, planning to raise the minimum LCR level to 92.5% by the end of December of 2023 and 100% by July of 2023. However, in October of 2022, the bond market was significantly impacted by the risk of default stemming from Gangwon Province’s Legoland project. Consequently, authorities postponed the normalization plan—which aimed to increase the LCR minimum requirement to 92.5% by the end of December of 2022—by six months, extending the deadline to June of 2023 (refer to Table 4). We expect such rules to boost demand for KTBs.

TABLE 4

TIMELINE OF THE INTRODUCTION AND ADJUSTMENT OF LCR REGULATIONS

Figure 4 highlights in yellow shading the gap banks needed to close to reach the final liquidity coverage ratio (LCR) target of 100%. While the majority of banks, on average, met the LCR requirements, the figure indicates that some banks still needed to purchase additional high-quality liquid assets, such as Korea Treasury bonds, to fully comply with the LCR regulation.

FIGURE 4.

DISTRIBUTION, MEAN, AND MEDIAN OF LCRS FOR COMMERCIAL BANKS (2015-2019)

Note: 1) ⓐ Announcement of the introduction of the liquidity coverage ratio (LCR) (August 27, 2014), ⓑ Revision announcement reducing the initial LCR requirement from 100% to 80% (December 24, 2014), ⓒ Implementation of the 80% LCR requirement (January 1, 2015), ⓓ Implementation of the 85% LCR requirement (January 1, 2016), ⓔ Implementation of the 90% LCR requirement (January 1, 2017), ⓕ Implementation of the 95% LCR requirement (January 1, 2018), ⓖ Implementation of the 100% LCR requirement (January 1, 2019); 2) Commercial banks include Kookmin, Shinhan, Woori, Hana, Standard Chartered Korea, Citibank Korea, Korea Exchange Bank (defunct), Kyongnam, Kwangju, Daegu, Busan, Jeonbuk, and Jeju Bank; 3) The light blue shaded area indicates the portion of banks with LCR ≥ 100%; 4) The yellow shaded area indicates the portion of banks with LCR < 100%; 5) The red dashed line represents the regulatory LCR threshold over time.

Source: Financial Supervisory Service, Financial Statistics Information System.

A. Model Specification: MF-DNS Model

Before introducing the Macro-Factor Dynamic Nelson-Siegel (MF-DNS) model, we briefly describe the standard Dynamic Nelson-Siegel (DNS) model, which does not consider macroeconomic variables. This introduction provides an understanding of yield curve modeling using latent factors and subsequently expands to the MF-DNS model incorporating macroeconomic variables.

In the basic DNS model, the yield curve at a given time is determined by three factors: the level, slope, and curvature, represented as follows:

In equation (1), y_t(τ_i) denotes the yield at time t on a bond with maturity τ_i, for i=1, ⋯, N. The factors l_t, s_t, and c_t denote the level, slope, and curvature factors, respectively.

The factor loadings for level, slope, and curvature are denoted as and are defined by

These factor loadings allow interpretations of l_t, s_t, and c_t as the level, slope, and curvature. The first factor loading is 1, indicating that the level factor uniformly affects all maturities, typically interpreted as the level factor. The second factor loading converges from 1 to 0 as the maturity τ increases, meaning it is larger for shorter maturities and is thus interpreted as the slope factor influencing the short-to-long-term interest rate differential. The speed at which this slope factor loading converges to 0 is governed by the decay parameter λ.

The third factor loading initially increases with the maturity and then declines toward zero, thus strongly influencing medium-term maturities and being interpreted as the curvature factor. The measurement error is assumed to follow a normal distribution, with a diagonal covariance matrix.

In the MF-DNS model, considering macroeconomic variables, the three yield curve factors (l_t, s_t, c_t) depend on their lagged values and contemporaneous macroeconomic variables m_t , represented as follows:

Here f_t=(l_t, s_t, c_t) and v_t follow a normal distribution N(0, Ω) without restrictions on the covariance matrix.

Equations (1) and (3) can be expressed in a state-space form as measurement and transition equations:

Measurement equation:

Transition equation:

The measurement error and the transition error are assumed to be independent. The impact of macroeconomic variables on current yields can thus be derived as shown below.

B. Model Estimation Method

The parameters for the estimation of the measurement equation (3) and transition equation (4) are μ, G, and Ω, collectively denoted as θ.

This study uses Bayesian estimation to estimate these parameters. Previous studies employing the MF-DNS model commonly use either maximum likelihood estimation (MLE) or Bayesian methods. With many exogenous macroeconomic variables in the MF-DNS model, the number of parameters can lead to severe local maxima issues in the MLE case. Therefore, Bayesian estimation is applied here.

The prior distributions are specified as follows. Each diagonal element of the measurement-error covariance matrix Σ follows an inverse gamma distribution. The intercept vector μ and coefficient matrix G follow normal distributions, whereas the transition-error covariance matrix Ω follows an inverse Wishart distribution. The prior hyperparameters follow Kim and Hong (2022), as detailed in Appendix Table C1.

C. MCMC Estimation Process

Given the initial parameter values, posterior distributions are sampled using Markov Chain Monte Carlo (MCMC) simulations. The burn-in period for the MCMC simulation consists of 100 iterations, followed by 1,000 posterior samplings. The MCMC steps are summarized below, with detailed steps provided in Appendix C:

A. Data

Figure 5 shows the trend in Korea Treasury bond (KTB) issuance yields by maturity at the time of the repo market regulations. Non-bank repo regulations began with an institutional improvement announcement on July 27, 2010, which restricted securities firms’ call borrowing to within 100% of their equity capital. These restrictions were progressively tightened over approximately five years, culminating in the complete exclusion of secondary financial institutions from the call market on February 24, 2015.

FIGURE 5.

KOREA TREASURY BOND (KTB) ISSUANCE YIELDS BY MATURITY (2010-2015)

Note: ① Announcement of reduction in securities firms’ call borrowing limit to within 100% of equity capital (July 27, 2010), ② Implementation of the 100% limit (October 1, 2010), ③ Announcement of the further reduction in call borrowing limit to within 25% of equity capital (June 22, 2011), ④ Implementation of the 25% limit (July 12, 2012), ⑤ Announcement of the complete exclusion of secondary financial institutions from the call market (November 22, 2013), ⑥ Implementation of the exclusion (February 24, 2015).

Source: Financial Supervisory Service, Financial Statistics Information System.

Meanwhile, Figure 6 illustrates the trend in KTB issuance yields by maturity at the time of the liquidity coverage ratio (LCR) regulations. The banking sector LCR regulation was first announced on August 27, 2014, initially setting the minimum LCR level at 100%. However, this requirement was relaxed to an 80% minimum level, effective from January 1, 2015. Subsequently, the minimum LCR level was increased annually by five percentage points until January 1, 2019. Additionally, the LCR requirements were partially relaxed in 2020 due to liquidity supply constraints resulting from the COVID-19 pandemic.

FIGURE 6.

KOREA TREASURY BOND (KTB) ISSUANCE YIELDS BY MATURITY (2015-2019)

Note: ⓐ Announcement of the liquidity coverage ratio (LCR) introduction (August 27, 2014), ⓑ Revision announcement reducing the initial LCR requirement from 100% to 80% (December 24, 2014), ⓒ Implementation of the 80% LCR requirement (January 1, 2015), ⓓ Implementation of the 85% LCR requirement (January 1, 2016), ⓔ Implementation of the 90% LCR requirement (January 1, 2017), ⓕ Implementation of the 95% LCR requirement (January 1, 2018), ⓖ Implementation of the 100% LCR requirement (January 1, 2019).

Source: Financial Supervisory Service, Financial Statistics Information System.

Figure 7 illustrates the monthly trends in Korea government bond issuance. As clearly depicted, prior to the onset of COVID-19 in 2019, the issuance of government bonds amounted to 101.5 trillion KRW. However, from 2020 onward, government bond issuance surged dramatically to 180.5 trillion KRW annually due to expansionary fiscal policies aimed at mitigating the economic impact of COVID-19. In this context, primary dealers (PDs) of government bonds were obligated to underwrite at least 10% of the tendered volume and had the right to purchase up to 30% of the tendered amount. In 2020, banks such as KB Kookmin Bank, IBK Industrial Bank, Nonghyup Bank, Hana Bank, and Standard Chartered Bank Korea served as PDs, inevitably increasing banks' holdings of government bonds.

FIGURE 7.

KOREA TREASURY BOND ISSUANCE (2015-2021)

Source: Ministry of Economy and Finance (Korea), Korea Treasury Bond Market Website.

In our empirical analysis, we include government bond issuance as a control variable in the analytical model to account for the effects of increased bond issuance and the corresponding rise in bond holdings by banks. The introduction of financial stability regulations affects liquidity demand, in turn affecting bond yields. Typically, an increase in the bond issuance volume indicates greater demand for funds, resulting in higher bond yields. This relationship was analyzed earlier in Kim and Hong (2022). Therefore, if bond issuance volumes were not controlled, the introduction of financial stability regulations, which increase the liquidity supply, may not reflect a corresponding decrease in bond yields. Consequently, the bond issuance volume is included as an additional control variable in this study.

Table 5 provides the basic descriptive statistics pertaining to the variables used in the MF-DNS analysis from 2010 to the end of 2019. To specifically analyze the impact of the LCR regulation announcement on interest rates, descriptive statistics were separated into pre-regulation (2010-end of 2014) and post-regulation (2015-end of 2019) periods. The total number of observations for the entire period (2010–2019) is 2,476, limited to business days. The number of observations for 30-year KTBs differs because issuance began on September 11, 2012. On average, KTB issuance yields increase with maturity.

TABLE 5

SUMMARY STATISTICS OF MF-DNS FACTORS

Source: Author’s calculations based on data from the Bank of Korea Economic Statistics System, Financial Supervisory Service Financial Statistics Information System, and Korea Securities Depository OTC Repo Statistics (as of December of 2016 and February of 2022).

Because KTB yields generally correlate with policy rate movements, this analysis also considered Korea and U.S. policy rates. Additional explanatory variables included the 10-year U.S. Treasury yield, commercial paper (CP) rates (91-day), call rates (overnight), consumer price index, industrial production index, and KOSPI index.

B. Effects of Restricting Securities Firms’ Call Market Participation

This study analyzes the impact of regulations restricting call borrowing by securities firms on Korea Treasury bond (KTB) yields, specifically examining how each stage of announcement and implementation affected yields. We estimated changes in bond yields across maturities around key announcement and implementation dates of the call borrowing restrictions, as presented in Figures 3 and 4. To achieve this, policy dummy variables were added to the Macro-Factor Dynamic Nelson–Siegel (MF-DNS) model alongside various macroeconomic variables. Additionally, to isolate the effects of the regulatory interventions from other potentially influential factors, we limited our analysis to a short event window of three months before and after each regulatory event.

The standard Dynamic Nelson–Siegel (DNS) model explains changes in yield curves solely based on three latent factors: the level, slope, and curvature. However, the MF-DNS model expands upon this by incorporating macroeconomic variables, allowing for analysis of yield curve shifts caused by these additional channels. When these macroeconomic variables include dummy variables representing specific events, it becomes possible to quantify their impact on the yield curve through changes in the level, slope, and curvature factors.

The regulations under analysis include the policy limiting securities firms' call borrowing to 25% of their equity capital (announced on June 22, 2011 and implemented on July 12, 2012) and the complete restriction on secondary financial institutions' participation in the call market (announced on November 22, 2013 and implemented on February 24, 2015). Prior to these, a 2010 regulation limited call borrowing to 100% of equity capital. However, because no securities firm exceeded the 100% limit at the time, this earlier regulation is not expected to have significantly impacted KTB demand and yields.

1. Analysis of the Impact of the 25% Call Borrowing Restriction on Securities Firms

This section presents the analysis of how limiting securities firms' call borrowing to 25% of their equity (the “25% Call Borrowing Restriction”) influenced the KTB yield curve. The analysis covers a period of three months before the regulation announcement and three months after its implementation. Dummy variables representing the period between the announcement (June 22, 2011) and implementation (July 12, 2012) and the period after implementation were incorporated into the MF-DNS model alongside macroeconomic variables.

a) MF-DNS Parameter Estimation Results

Table 6 presents the parameter estimation results of the MF-DNS model, including macroeconomic variables and the KTB issuance volume, to analyze the effects of the 25% call borrowing restriction. It is noteworthy that these results reflect the model controlling for the issuance volume. In contrast, the estimation results of the model without controlling for the KTB issuance volume are provided in Appendix Table A3. Both models contain identical macroeconomic variables aside from the inclusion of the bond issuance volume, and both tables provide mean parameter estimates and posterior distributions..

TABLE 6

SUMMARY MF-DNS PARAMETER ESTIMATES: INTEREST RATE CHANGES UNDER THE 25% CALL BORROWING LIMIT (RELATIVE TO SECURITIES FIRMS’ EQUITY) WITH ISSUANCE VOLUME CONTROL

Results considering bond issuance (Table 6) indicate that the announcement and implementation of the 25% call borrowing restriction primarily influenced the slope and level factors of the yield curve. Specifically, the estimated coefficient for the slope factor at the announcement stage is 0.0117, while the estimated coefficient for the level factor at the implementation stage is -0.0878. Both coefficients are statistically significant. Estimates excluding bond issuance also display similar statistical significance (Appendix Table A3). The slope coefficient at announcement (0.0127) is larger than the coefficient excluding bond issuance (0.0117), whereas the level coefficient at implementation (-0.0856) is slightly smaller than that without bond issuance control (-0.0878).

According to equation (6), the impact of an event on bond yields according to the maturity date is determined by multiplying the factor loadings by the G matrix. Thus, despite differences in signs among the level, slope, and curvature estimates, the direction of their combined impact on yields can remain consistent. For example, considering the 3-year KTB (τ = 36 months), using the decay parameter (λ) of 0.0609 from previous studies, the factor loadings for the slope and curvature factors are 0.405 and 0.293, respectively, while the level factor loading is 1.

Multiplying these loadings by the estimated coefficients from Table 6 provides the impacts on the 3-year KTB yield, showing values of -0.0878 (level), 0.0016 (slope), and 0.0030 (curvature). Thus, the 25% call borrowing restriction reduced the 3-year bond yield primarily through the level factor, whose absolute impact was significantly greater than those of the slope and curvature factors. Combining these effects, the total estimated impact on the 3-year bond yield at the time of implementation is -0.0832, consistent with the estimates in Table 7.

TABLE 7

MF-DNS ANALYSIS RESULTS: INTEREST RATE CHANGES UNDER THE 25% CALL BORROWING LIMIT (RELATIVE TO SECURITIES FIRMS’ EQUITY) WITH ISSUANCE VOLUME CONTROL

Note: 1) The data analysis period spans the dates of April 27, 2011 to August 10, 2012; 2) The 25% call borrowing limit regulation was announced on May 27, 2011 and implemented on July 12, 2012.

b) Effects of the 25% Call Borrowing Restriction on Bond Yields by Maturity

Table 7 presents the results of the analysis of the 25% call borrowing restriction's impact on yields across maturities, revealing that limiting securities firms' call borrowing to 25% of their equity substantially reduced KTB yields, with the impact more pronounced during the implementation phase. As previously noted, the estimation results without controlling for the KTB issuance volume are provided in Appendix Tables A1 and A2.

Specifically, implementation of the 25% call borrowing restriction lowered yields across all maturities by approximately 8.24–8.69 basis points, with these results being statistically significant as well (5–95% credible interval). Meanwhile, the announcement phase resulted in a relatively small reduction of about 0.48–1.08 basis points, with weaker statistical significance compared to the implementation phase.

In total, restricting securities firms' call borrowing to 25% of equity capital cumulatively decreased bond yields by approximately 8.72–9.77 basis points. Notably, the reduction in yields was more significant during implementation compared to the time of announcement, suggesting a delayed shift from call borrowing to repo-based funding, characteristic of short-term funding market dynamics.

The difference-in-differences (DiD) analysis results in Table 2 and 3 reveal that the implementation of the 25% call borrowing limit, relative to securities firms' equity, resulted in an increase of approximately KRW 5 trillion in securities firms' fund-raising through repo (RP) transactions over two years. Given that approximately 65% of RP transactions are collateralized by KTBs, the increased KTB demand specifically due to collateralization needs can be estimated at approximately KRW 3.25 trillion (KRW 5 trillion × 0.65). Thus, an increase in KTB supply of KRW 1 trillion corresponds to an estimated reduction in KTB yields by approximately 2.7 to 3.0 basis points (bp). This result aligns closely with findings in Kim and Hong (2022), where a KRW 1 trillion increase in KTB issuance (representing demand) was associated with an approximate rise of 3.2 to 4.3 bp in KTB yields. These findings provide insights into the elasticity of KTB yields in response to changes in market supply and demand.

2. Analysis of the Complete Restriction on Securities Firms’ Call Market Participation

We also examine the effects of completely restricting secondary financial institutions’ participation in the call market (the “Complete Call Market Restriction”) on KTB yields. The analysis period encompasses three months before the announcement and three months after the implementation. Dummy variables indicating the period between the announcement (November 22, 2013) and the implementation (February 24, 2015), and the period after implementation, were included in the MF-DNS model with macroeconomic variables to assess their effects on bond yields. As in the previous section, detailed MF-DNS parameter estimates with issuance-volume control are reported in Appendix Table A5.

Table 8 presents the analysis results of the impact of the “Complete Call Market Restriction” on bond yields by maturity. These results indicate that both the announcement and the implementation of this restriction significantly reduced bond yields. Specifically, the announcement phase reduced yields by approximately 1.17–1.88 basis points across all maturities, showing statistical significance (5–95% credible interval). Meanwhile, the implementation phase led to an additional yield decline in the approximate range of 0.68–0.82 basis points, also statistically significant.

TABLE 8

MF-DNS ANALYSIS RESULTS: INTEREST RATE CHANGES BEFORE AND AFTER THE ANNOUNCEMENT AND IMPLEMENTATION OF THE PRINCIPLED EXCLUSION OF SECURITIES FIRMS FROM THE CALL MARKET WITH ISSUANCE VOLUME CONTROL

Note: 1) The data analysis period spans the dates of December 28, 2012 to March 24, 2015; 2) The principled exclusion of securities firms from participation in the call market was announced on November 22, 2013 and implemented on February 24, 2015.

In total, the Complete Call Market Restriction cumulatively decreased bond yields by approximately 2.03–2.7 basis points. This reduction was notably larger during implementation compared to the corresponding announcement, mirroring earlier findings regarding the 25% call borrowing restriction. This suggests that securities firms gradually transitioned from call borrowing to repo-based funding, thereby influencing bond yields more substantially at the time of actual regulatory implementation.

C. Effects of LCR Introduction on Banks

We also examine the impact of three key events related to the introduction and initial implementation of the liquidity coverage ratio (LCR) regulation in 2014-2015: (1) the announcement of the LCR implementation (August 27, 2014), (2) the approval of revisions to banking supervisory regulations (December 24, 2014), and (3) the enactment of the LCR regulation (January 1, 2015). Specifically, the changes in Korea Treasury bond yields relative to the period before the LCR announcement (August 27, 2014) were estimated for each event. Dummy variables representing each event were added to the Macro-Factor Dynamic Nelson–Siegel (MF-DNS) model along with macroeconomic variables to isolate the impact of the regulatory announcements on bond yields. To control for external factors beyond regulatory changes, the analysis period was limited to within three months before and after each regulatory event. As in the previous sections, detailed MF-DNS parameter estimates with issuance-volume control are reported in Appendix Table A7.

Table 9 summarizes the estimated impact of the three LCR-related events—the announcement of the LCR implementation (August 27, 2014), the approval of revisions to banking supervisory regulations (December 24, 2014), and the enactment of the LCR regulation (January 1, 2015)—on Korea Treasury bond yields across maturities. The analysis shows that all three events consistently decreased yields across all maturities. Specifically, bond yields decreased by approximately 1.97–2.33 basis points following the announcement of LCR implementation, by 3.00–3.09 basis points after the approval of the banking supervisory regulation revisions, and by 1.15–1.92 basis points upon LCR implementation. These figures represent shifts in the demand curve for bonds after controlling for the bond issuance volume and other macroeconomic variables.

TABLE 9

MF-DNS ANALYSIS RESULTS: INTEREST RATE CHANGES BEFORE AND AFTER THE ANNOUNCEMENT AND IMPLEMENTATION OF THE LCR REGULATION WITH ISSUANCE VOLUME CONTROL

Note: The data analysis period spans May 28, 2014, to January 30, 2015.

The analysis also indicates that the shorter the bond maturity, the greater the yield reduction following the LCR regulation announcement. For example, 2-year and 3-year bond yields decreased by 2.33 and 2.29 basis points, respectively, greater reductions compared to the 10-year bond yield decline (2.06 basis points). In comparison, yields on 20-year and 30-year bonds decreased by approximately 1.99 and 1.97 basis points, respectively, indicating smaller differences compared to the 10-year bond.

This suggests that the LCR introduction potentially increased the demand for shorter-term bonds (maturing in less than five years) more than longer-term bonds (ten years or more). Due to their shorter-term liabilities (primarily deposits), banks have greater incentives to hold shorter-maturity treasury bonds for asset-liability management (ALM) purposes.