Electronic copy available at: https://ssrn.com/abstract=3015701 1 Market Segmentation and Limits to Arbitrage under Negative Interest Rates: Evidence from the Bank of Japan’s QQE * Takahiro Hattori Ministry of Finance Japan, Hitotsubashi University This version 2017/8 Abstract This paper decomposes the bond yield into the segmentation factor using a unique Japanese dataset. To control for the channel of future expectations and the term premium, we take advantage of data on the government guaranteed bond, which is an asset identical to the government bond, except for liquidity, and the fact that it has not been institutionally affected by demand from the Bank of Japan. We extend the model of Krishnamurthy et al. (2015) to capture the liquidity factor explicitly. Our result shows that the market has segmented during the time when the government bond yield became negative, although the segmentation factor is small during ‘normal’ times. JEL codes: E43, E52, E58, E65, G12, G14 Keywords: Preferred Habitat, Market Segmentation, Quantitative Easing, Term Structure of Interest Rate, Zero Lower Bound * The author would like to thank Junko Koeda, Masazumi Hattori, Etsuro Shioji, Toshiaki Watanabe, Takashi Unayama and seminar participants at Keio University and Summer Workshop on Economic Theory. The views expressed in this paper are those of the author and not those of the Ministry of Finance or the Policy Research Institute.
26
Embed
Market Segmentation and Limit to Arbitrage under Negative Interest … · Market Segmentation and Limits to Arbitrage under Negative Interest Rates: Evidence from the Bank of Japan’s
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Electronic copy available at: https://ssrn.com/abstract=3015701
1
Market Segmentation and Limits to Arbitrage under Negative
Interest Rates: Evidence from the Bank of Japan’s QQE*
Takahiro Hattori
Ministry of Finance Japan, Hitotsubashi University
This version 2017/8
Abstract
This paper decomposes the bond yield into the segmentation factor using a unique
Japanese dataset. To control for the channel of future expectations and the term
premium, we take advantage of data on the government guaranteed bond, which is an
asset identical to the government bond, except for liquidity, and the fact that it has not
been institutionally affected by demand from the Bank of Japan. We extend the model
of Krishnamurthy et al. (2015) to capture the liquidity factor explicitly. Our result shows
that the market has segmented during the time when the government bond yield became
negative, although the segmentation factor is small during ‘normal’ times.
JEL codes: E43, E52, E58, E65, G12, G14
Keywords: Preferred Habitat, Market Segmentation, Quantitative Easing, Term
Structure of Interest Rate, Zero Lower Bound
* The author would like to thank Junko Koeda, Masazumi Hattori, Etsuro Shioji,
Toshiaki Watanabe, Takashi Unayama and seminar participants at Keio University and
Summer Workshop on Economic Theory. The views expressed in this paper are those
of the author and not those of the Ministry of Finance or the Policy Research Institute.
Electronic copy available at: https://ssrn.com/abstract=3015701
2
1. Introduction
How does unconventional monetary policy affect the term structure of the interest
rate? According to the preferred habitat theory, the supply and demand for bonds
determine the yield curve in which the markets are segmented. A number of studies
attempt to test this hypothesis and measure the supply and demand effect by analyzing a
specific policy event. However, it is still not easy to identify how much the
demand/supply factor (the “segmentation factor”) directly affects the yield curve
because quantitative easing (QE) could work through several channels, including
affecting future expectations (“signaling”) and the term premium (“duration risk”).
In this paper, we utilize unique data from Japan to detect how the segmentation factor
affects the government bond yield. In April 2013, the Bank of Japan (BOJ) introduced
Quantitative and Qualitative Easing (QQE), with the aim of achieving a 2% inflation
rate. Under the QQE, the BOJ has purchased Japanese Government Bonds (JGBs) at an
annual pace of 80 trillion yen, suggesting that the QQE is an event that significantly
increases the demand for JGBs. On the other hand, the QQE policy has also impacted
on the yield curve through the signaling and duration risk channels by strengthening its
forward guidance, pledging to continue the QQE policy until inflation is stable above
the 2% target. In general, separating the segmentation factor from the other channels is a
challenging task and, although many studies attempt to estimate the effect of QE, only a
few papers explicitly decompose these factors.
Our identification strategy for separating the segmentation factor is to take the spread
of Japanese Government Guaranteed Bonds (JGGBs) and JGBs, relying on two
characteristics of these bonds. First, as Krishnamurthy et al. (2015) point out, it is
natural to consider that the signaling and duration risk channels should lower not only
3
JGB yields but also JGGB yields in the same way, such that taking the spread of JGBs
and JGGBs enables us to cancel out these two components. Second, the important
institutional feature of QQE is that the BOJ does not purchase JGGBs. This means that
a demand shock caused by the BOJ only affects JGBs, not JGGBs. We note that JGGBs
are an identical asset to JGBs except in relation to liquidity. Exploiting this clear
characteristic as a laboratory, we decompose the spread into the segmentation and
liquidity factors.
Our result shows that the arbitrage between JGBs and JGGBs should be sufficient
during times when the JGB yields are positive, suggesting that the segmentation factor
has a relatively small effect on determination of the yield curve. However, the
importance of the segmentation factor is drastically amplified when the JGB yields
become negative. Intuitively, although the purchases by the BOJ push the JGB yields
deep into negative territory, there are no investors for JGGBs if the yields hit the zero
lower bound (ZLB). Figure 1 shows the time series for the spread of JGGBs and JGBs,
indicating that only the JGB yields have become negative, whereas the JGGB yields hit
the ZLB. This suggests that the markets clearly exhibit segmentation among JGBs and
JGGBs. This result is consistent with the theoretical implications proposed by Vayanos
and Vila (2009) and Greenwood and Vayanos (2014). In these papers, the arbitrageurs
absorb shocks to the demand and supply of a specific maturity’s bonds and, if the
arbitrageurs are confronted with a large limitation to arbitrage, the term structure of the
bond shows extreme segmentation.
The mechanism of the market segmentation between JGBs and JGGBs under ZLB is
quite simple. The JGB yield can move into negative territory because the BOJ
4
purchases JGBs with a negative yield.1 From the investors’ points of view, even if they
hold JGBs with negative yields, they can earn the zero yield (or even a positive return)
as long as the BOJ purchases JGBs with a negative yield. On the other hand, JGGBs
have not been purchased by the BOJ and, therefore, investors have no incentive to buy
JGGBs that yield an interest rate of less than zero.
Our paper is related to several strands of literature that focus on the segmentation
factor using spreads. Krishnamurthy and Vissing-Jorgensen (2011) study the Federal
Reserve quantitative easing policies during 2008–11, using the spread between the very
highest quality corporate bonds and comparable maturity government debt.
Krishnamurthy and Vissing-Jorgensen (2012) find that the higher is the supply of
government debt, the larger is the spread of the very highest quality corporate bonds.
Badoer and James (2016) present results that are consistent with Krishnamurthy and
Vissing-Jorgensen (2011, 2012). Krishnamurthy et al. (2015) focus on the spread
between the OIS swap rate and government bond yields to evaluate the European
Central Bank’s policy.
Our paper contributes to the existing literature in the following ways. First, this paper
is the first to describe how the ZLB affects the term structure in the context of the
preferred habitat theory. Numerous studies, including Dai et al. (2007), Koeda (2013),
and Christensen (2015), use regime switch models to examine the dynamics of bond
yields, including ZLB. In contrast to these studies, in our paper, identification is
achieved by using the institutional information of QQE, under which the BOJ only
purchases JGBs (not JGGBs), and we provide an interpretation of the regime shift in the
context of the preferred habitat theory.
1 The BOJ purchased JGBs with negative yields for the first time on September 9, 2014. This
preceded the implementation of the QQE with a Negative Interest Rate policy.
5
Second, compared with the previous studies, we utilize an asset that is more ideal for
analysis, the government guaranteed bond. For example, Krishnamurthy and
Vissing-Jorgensen (2011, 2012) and Krishnamurthy et al. (2015) are based on a
significant nondefault component, although high rated bonds and the OIS swap rate
could contain a slight default risk or a counterparty risk. To overcome this problem, we
use the JGGB, which has the same credit risk as the JGB and nonderivative assets. In
addition, our paper extends the existing literature by for the first time explicitly dealing
with the liquidity factor, which is nonnegligible for the asset pricing.
Third, we construct a liquidity measure using the spread of JGGBs and JGBs,
controlling for the segmentation factor. The liquidity measure is available publically
through our website (https://sites.google.com/site/hattori0819/data). Although there are
many studies that construct the liquidity measure by interpreting the spread of
government guaranteed bonds and government bonds, this is the first paper to explicitly
control the segmentation factor in this literature.
The remainder of the paper is organized as follows. Section 2 describes the features of
QQE and JGGBs. Section 3 contains the model and estimation strategy and Section 4
describes the data. Section 5 presents the results and the robustness check and Section 6
concludes.
2. Description of QQE and JGGBs
2.1 Government guaranteed bonds as a liquidity premium under positive JGB yields
Incorporated administrative agencies run businesses for public purposes in Japan in
their role as government agencies. The central government guarantees their debt up to a
maximum amount provided in the budget. The incorporated administrative agencies
issue bonds and the central government explicitly and fully guarantees these bonds. In