Essays on European bond markets

This dissertation focused on a number of issues that are of importance in the current
European bond market. In the past years, the fiscal policy of the Eurozone members,
advances in the technology of trading platforms and the introduction of a single
currency have reshaped the fixed income markets in Europe. These developments
have resulted in a far going integration of Eurozone capital markets. Moreover, the
massive amounts of debt issued by Japan and the United States combined with
the deteriorating stock markets in 2000-2002 have also resulted in an increased
asset allocation to European bonds. Currently, the Eurozone bond market is the
third largest market in the world with the German Schatze, Bobl and Bund futures
among the heaviest traded financial contracts. In recent years, the empirical work
on the microstructure of financial markets has received considerable attention in the
academic literature. Most of the substantial empirical work in this area pertains
to stock markets. Given the emphasis on stock markets in the theory and the
availability of data, this is understandable. On the other hand, in terms of both
capitalization and trading volume, bond markets are bigger than stock markets.
Research on bond markets is also interesting because of their special structure.
These markets are centered around a large number of professional dealers. Outside
customers trade with the dealer of their choice. Volume is high and interdealer
trading is frequently being observed. This work could be divided into two parts.
The first part focused on the microstructure of European bond markets. The second
part focused on the growing European inflation-linked bond market. Notice that
throughout this dissertation, we distinguished between market participants who can
set new prices (market makers or dealers) and market participants that can only
take prices (traders).
In the first part, we turned our attention to the microstructure of Eurozone bond
markets by studying the Belgian. French. German, and Italian bonds. We started
with a review of microstructure literature in chapter 2 and studied inventory and
information models. For inventory models, the equilibrium price formed by dealers
compensates for non-optimal inventory positions while the equilibrium price in information
models protects against adverse selection. Because a competitive market
structure prevails in the Eurozone bond market, interdealer trading is important.
We therefore provided a discussion about interdealer trading and its impact on the
formation of prices. We showed that in these interdealer models, inventory and
information asymmetry are still important because order flows are sources of private
information. This private information, combined with risk-averse speculation
among dealers, will result in strategic interdealer behavior. We ended chapter 2
with a discussion about some variables that could be observed by an econometrician
and therefore useful for testing market microstructure topics. We turned our attention
to the bid-ask spread, price volatility and order flow. The bid-ask spread is
an important instrument for dealers because it controls the incoming and outgoing
order flow, it compensates dealers for their market making activity and it serves as
a.. protection against adverse selection. Based on quote data, one can calculate the
average,, effective and realized spread. Based on transaction data, one can calculate
implicit trading costs. An important role is also put aside for price volatility. According
to theory, the more informed traders we have, the more price volatility will
occur as market makers will change their quotes more often to protect themselves
against adverse selection. The empirical papers indeed support this idea and show that the arrival of information is an important contributor to intraday price volatility--
Monitoring order flow is also crucial in financial markets. Even if markets are
driven by publicly available news, there exist information asymmetry in the form of
private order flow as it can observed only by the parties involved in the transaction.
In addition, the impact of order flow is time-varying. Not only does it depend on
the size and its direction, but also on the arrival of information in the market.
In chapter 3, we turned our attention to the trading activity of the MTS Global
Market system, which is the most important European interdealer fixed income
trading system. This system is composed of a number of trading platforms on
which designated bonds can be traded. The trading system is fully automated and
effectively works as an electronic limit order market. The first interesting feature
of the MTS trading platform is its organizational setup. Fixed income securities
can be traded on a domestic and an European (EuroMTS) platform. The range of
securities being traded on the domestic platform is however much larger than on the
EuroMTS trading platform. A dealer on the domestic trading platform can therefore
trade a much wider range of bonds. However, the existence of both trading platforms
suggests differences and we therefore asked ourselves the following question: Why
would a market maker with entrance to the local platforms also operate on the
EuroMTS trading platform? To answer this question, a detailed study on the costs
and the dynamics of price formation is needed. Throughout the paper, we provided
a comparison of the trading costs and price dynamics on the domestic MTS markets
and the EuroMTS by calculating comparative measures of liquidity, such as the
quoted and effective spread. We showed that despite the apparent fragmentation of
trading on domestic platforms and EuroMTS. the markets are closely connected in
terms of liquidity. The second interesting feature of the MTS Global Market system
is its pure interdealer platform. This allowed us to study the price and order flow
dynamics under competitive market making. We asked ourselves: What are the
dynamics of prices in the Eurozone fixed income market under competitive market
making and interdealer trading'.'' What is the role of trading intensity? Interestingly.
the study of trading intensity and its relation to price and order flow dynamics do
not explicitly take the role of interdealer trading into consideration. Single dealer
models argue that there exist a positive relation between information and trading
intensity as more informed traders arc active during large market activity. This
means that any unexpected trade during active trading has a higher impact on
prices.. Some authors also documented the same empirical results. We think that
interdealer trading may shed a different light on these results. Dealers use interdealer
trading to control their inventory position but interdealer trading is more costly
compared to outside customer trading because dealers have to pay a fee (to the
otherr dealer) rather than receiving a fee (from an outside customer). In addition,
there is less need to adjust the spread as traders enter the market on a frequent basis,
which by definition occur during high market activity. Next to these searching costs,
interdealer trading can also result in a repeated passing of inventory among dealers as
they have the moral obligation to quote prices. The repeated passing of inventory is
called "hot-potato" trading and this creates noise in the pricing process when dealers
are risk averse and speculative. In other words, a dealer's decision to conduct an
interdealer trade depends on his ability to offset his inventory using customer order
flow.. Not only is this cheaper compared to interdealer trading; it also avoids the
"hot-potato"" process. This implies that the price impact of a trade is much larger
during periods where trading intensity is low because customer flows are scarce.
Our data provided an opportunity to test this. We documented the following: order
flows are strongly correlated but the correlation gradually decreases over time. We
also found the impact of a trade in a low trading intensive environment having a
larger impact on price than in a high trading intensive environment. This contrasts
the findings for stock markets but confirms the dealer's preference to trade with
outside customers rather than the more costly interdealer trading. When we took
announcement effects into consideration, we found the overall price sensitivity to
order flow being stronger and this reflects the information asymmetry in order flow.
However,, this price sensitivity is magnified when trading activity is low. The latter too calculate the inflation premium within a break-even framework, expected inflation
extracted from survey data is incorporated. The inflation premium is then the
difference between the nominal yield, real yields and the expected inflation. Still,
the proposed method assumes that the index-linked curve estimated from inflationlinked
price data is equal to the real term structure of interest rates. There is
however a difference due to liquidity and imperfect indexation. Most papers focusing
on the liquidity premium do understand the importance of liquidity but do not
explicitly model liquidity as a driving factor of bond prices. It is well known however
that liquid securities are traded at a premium. Taking liquidity into account
allowed us to use data on both nominal and inflation-linked debt in markets where
the outstanding amount of inflation linked debt is small. In France for example,
the fraction of inflation-linked bonds stands at 7% of total debt while the Italian
treasury estimates that some 1.3% of its current debt is inflation-linked. The US
treasury market is the worlds largest inflation-linked bond market and the outstanding
amount of treasury inflation protected securities (or TIPS) equals 150 billion.
This is approximately 6% of the total outstanding tradable US treasury debt. To
some extent, the problem of liquidity can be avoided by analyzing the UK bond
market because almost 25% of their tradable debt is inflation linked. The objective
chapter 5 is to estimate the inflation premium by taking liquidity into account. This
allowed us to study the empirical properties of the term structure of real rates in
the Eurozone bond market. We used data from French index-linked and nominal
bonds and estimate the inflation and liquidity premium in a state space framework
using the extended Kalman filter and quasi-maximum likelihood. In particular, we
simultaneously derived the nominal and real term structure of interest rates and this
enabled us to calculate the price of any discount bond. In order to fit coupon bearing
bonds into an affine structure, we used the extended Kalman filter to linearize
the state equations. A state space approach is also useful for taking advantage of
cross-sectionall and timeseries behavior of nominal and real rates and this should
reduce the instability of timeseries. Our findings are as follows: in the absence of
liquidity, the inflation premium runs from 113 basis points to some 250 basis points
across the curve. These numbers implies that the inflation premium in long-term
bonds is more than 2 times larger compared with the short-end of the curve and
comparable with US TIPS. The liquidity premium in real bonds equals some 6 basis
points for bonds maturing in 2009 and is slightly humped shaped with a peak at the
10-year bond. If liquidity is taken into consideration, the expected nominal yield
spread between nominal and real bonds equals some 15 basis points in the short end
of the curve but increases to 135 basis points in the long end of the curve. For shortterm
(long-term) bonds, the liquidity premium accounts for some 5% (10%) of the
total real risk premium. On the other hand, the inflation premium is a prominent
factor in nominal bonds as it account for more than 50% of the total risk premium
across the term structure for nominal interest rates. As a final remark, although
the contribution of the liquidity premium to total premium is small, it has a large
impact on the expected nominal yield spread through the expected liquidity level.
We find that this yield spread is upward sloping as it runs from 76 basis points to
some 198 basis points for the 2032 bond.