Attachment
Attachment
Attachment
Financial System
CGFS Papers
No 55
Fixed income market
liquidity
Report submitted by a Study Group established by the
Committee on the Global Financial System
The Group was chaired by Denis Beau (Bank of France)
January 2016
© Bank for International Settlements 2016. All rights reserved. Brief excerpts may be
reproduced or translated provided the source is stated.
Fixed income market liquidity plays a crucial role in the conduct of monetary policy
and the stability of the financial system. Therefore, central banks have a vital interest
in monitoring liquidity conditions as well as the drivers that affect their robustness
during episodes of market stress.
The Committee on the Global Financial System (CGFS) monitors structural
developments in fixed income markets on a regular basis. In November 2014, it
published a report entitled Market-making and proprietary trading: industry trends,
drivers and policy implications (CGFS Publications, no 52). Recent market events
argued in favour of updating the initial assessment in that report, and of expanding
the analysis to evaluate the robustness of market liquidity and its underlying drivers.
To pursue this effort, the CGFS decided to reconvene the original Study Group
under the chairmanship of Denis Beau (Bank of France).
The following report summarises the Group’s main findings. It highlights that
fixed income markets are in a state of transition. Dealers have continued to cut back
their market-making capacity in many jurisdictions. Demand for market-making
services, in turn, continues to grow. The effects of these diverging trends have, thus
far, not manifested themselves in the price of immediacy services, but rather they
are reflected in possibly increasingly fragile liquidity conditions. Key drivers of
current trends in liquidity include the expansion of electronic trading, dealer
deleveraging, arguably reinforced by regulatory reform, and unconventional
monetary policies. Given the transitional state of fixed income markets, regulators
appear to be facing a short-term trade-off between less risk-taking by banks and
more resilient market liquidity. Yet, in the medium term, measures to bolster market
intermediaries’ risk-absorption capacity will strengthen systemic stability, including
through a more sustainable supply of immediacy services. To help ensure a smooth
transition, the report argues for a close monitoring of liquidity conditions as well as
an ongoing assessment of how new liquidity providers and trading platforms are
affecting the distribution of risks among market participants.
I hope that this work can contribute to the ongoing discussion about the
changing nature of fixed income market liquidity, and can serve as a resource for
policymakers as well as market practitioners interested in the broader implications
of these changes.
William C Dudley
Chairman, Committee on the Global Financial System
President, Federal Reserve Bank of New York
1. Introduction ....................................................................................................................................... 4
3. Drivers ................................................................................................................................................ 15
Technology and competition: shaping adjustments in trading and
business models .................................................................................................................. 15
Bank deleveraging and regulation ........................................................................................ 16
Monetary policy ............................................................................................................................ 20
4. A new regime for fixed income market liquidity? – Market and policy
implications ............................................................................................................................. 22
Market implications ..................................................................................................................... 22
Strengthening market resilience ............................................................................................ 23
Adapting policy to the new market environment ........................................................... 24
References ................................................................................................................................................ 26
Reduced supply vs rising demand. Overall, the main trends identified by the
CGFS’s initial assessment remain in action. Dealers have continued to lower their
market-making capacity and willingness in many jurisdictions, focusing on activities
that require less capital. Demand for market-making services, in turn, continues to
grow given the expansion of primary bond markets and increased bond holdings by
market participants who rely on dealers’ immediacy services (eg asset managers).
The impact of diverging trends in liquidity supply and demand differs across bond
markets.
Liquidity increasingly fragile in benchmark bond markets? For benchmark
sovereign bonds, liquidity appears little changed, judging by a variety of market-
based metrics. Some signs do, however, point to greater fragility in liquidity
conditions. While it is difficult to identify the drivers of this fragility (see below),
recent episodes of market stress suggest that, in some markets, the rise of
algorithmic trading may have had an impact on liquidity.
Liquidity bifurcation continues. For other markets, such as those for off-the-run
sovereign bonds and corporate bonds, there is evidence of bifurcation, with liquidity
deteriorating most in those market segments that have historically been less deep
than others. In these segments, the reduction in dealers’ market-making capacity
seems to have had a greater impact on liquidity, given the limited availability of
substitutes to their services.
Adjustment mainly through quantities, not prices. The above trends potentially
imply upward pressure on trading costs and, ultimately, higher costs of financing in
primary markets. Yet, price-based metrics of these costs, such as bid-ask spreads
and liquidity premia, provide little evidence of any significant changes thus far. One
reason, as suggested by many market participants, is that the main margin of
adjustment is through quantities, rather than prices. The trading of large amounts,
for example, has reportedly become more complex and time-consuming, as many
dealers are reluctant to warehouse large positions. Another reason is conjunctural
factors (see below) that affect current measures of market liquidity.
Fixed income market liquidity plays a crucial role in the conduct of monetary policy
and the stability of the financial system. Therefore, central banks have a vital interest
in monitoring liquidity conditions as well as the drivers that affect their robustness
during episodes of market stress.1
Background. In November 2014, the Committee on the Global Financial System
(CGFS) published a report on the current trends and drivers of market-making and
proprietary trading in fixed income markets (CGFS (2014)). The report pointed to
signs of increased liquidity bifurcation and fragility, with market activity
concentrating in the most liquid instruments and deteriorating in the less liquid
ones. Several factors, including both structural and conjunctural ones, were found to
be driving these developments.
The report concluded that, in future, the pricing of immediacy services could
become more consistent with actual market-making capacity and costs. Yet, given
that markets were in a state of transition, it remained difficult to provide a definitive
overall assessment. Against this background and given recent market events,2 the
CGFS decided to reconvene the original Study Group under the chairmanship of
Denis Beau (Bank of France) in order to update its initial assessment and to engage
in follow-up analysis on trends in fixed income market liquidity.
A plethora of markets. The term “fixed income” spans a wide range of markets
that differ along many dimensions. One such dimension is market liquidity, with the
most liquid markets (eg those for major sovereign bond futures) trading almost
exclusively on electronic platforms, often on central limit order books, and where
liquidity is provided by a multitude of market participants. These include the
“traditional market-makers” (eg banks and securities firms) and, to an increasing
extent, other less-regulated entities (eg principal trading firms; “PTFs”) that are
challenging their peers’ business models.
On the other end of the spectrum are markets with infrequent and thin trading,
such as many corporate bond markets, where different features of the underlying
securities hamper the matching of demand and supply.3 These markets have
broadly maintained a dealer-centric setup, where traditional market-makers provide
immediacy services by warehousing assets, often over longer periods of time, to
meet client orders (principal-based trading) or, acting as brokers, seek to match
investors willing to buy and sell (agency-based trading).
Given these differences, the effects of structural changes (eg technological
innovations, regulation) and conjunctural developments (eg monetary policy) are
1
See, for example, CGFS (1999) for a more detailed discussion of the link between monetary policy
and fixed income market liquidity.
2
For two recent examples, see the case studies presented in this report (Box 2 and 3).
3
Given the large number of corporate bonds, often with relatively small total outstanding amounts
per issue and a variety of contractual features (eg call options), the probability of finding matches in
investor supply and demand for any given bond is much more limited than for other asset classes
(eg equities, sovereign bonds). Trading these bonds, therefore, typically relies on intermediation by
market-makers. For related analysis, see eg Goldstein et al (2007) or Bao et al (2011), and CGFS
(2014) for a discussion.
4
For a discussion of the concept, see CGFS (1999); for a discussion of suitable measures see
CGFS (2014), in particular Appendix 3.
United States2 Japan3 and euro area4 Korea and Mexico5 China and India6
256th of a point Bps Bps KRW per 10,000 par value Bps Bps
4.5 60 3 3 3 90 180
3.0 40 2 2 2 60 120
1.5 20 1 1 1 30 60
0.0 0 0 0 0 0 0
2006 2009 2012 2015 2011 2013 2015 2013 2014 2015 2013 2014 2015
Two-year 10-year Lhs: Rhs: Korea (lhs) China (lhs) India (rhs)
Italy Japan Mexico (rhs)
Germany
1
Monthly averages. 2 Spreads for US Treasury notes in the inter-dealer market; these spreads are reported in 256ths of a point,
3
equivalent to about 0.39 cents per US$ 100 face value (par). 10-year Japanese Government bonds. 4 Medium-term government
5
bonds (BTPs) for Italy, 10-year government bonds for Germany. Five-year Korean and Mexican government bonds. 6 Most liquid
benchmark sovereign bonds.
5
The cut-off date for the Group’s initial assessment was approximately end-2013; see CGFS (2014).
6
Quoted depth serves as a proxy of market depth. It measures the quantity of securities that is bid
(or offered) at the posted bid and ask prices.
Quoted depth1 Average transaction size2 Price impact: US Treasuries Price impact: Italian BTPs
EUR bn USD bn Millions in local currency 32nd of a point Bps
12 4 16 9 80
10 3 12 6 0
8 2 8 3 –80
6 1 4 0 –160
4 0 0 –3 –240
11 12 13 14 15 11 12 13 14 15 11 12 13 14 15 11 12 13 14 15
United States (rhs) United States Spain 2-year 10-year Impact on: bid ask
Italy (lhs) Italy 5-year
1
Quoted depth at five levels for two-year US Treasury notes; quoted depth for the five best quotes exhibited in MTS Cash for medium- and
long-term Italian government bonds (BTPs); monthly averages. 2 Average transaction size for two-year Treasury notes (United States); for
a weighted average of all Italian sovereign bonds; and for Spanish public debt; three-month moving averages. 3 Price change per $1
billion net order flow; monthly averages. 4 Estimated impact of high-value orders (buy and sell orders of €50 million) on the quoted
prices of benchmark 10-year BTPs from January 2010 to the end of June 2015.
7
Similar developments are also found by Kurosaki et al (2015) for JGB futures markets.
8
For some markets, alternative measures of dealer positions are available that can complement the
analysis (data not shown). If dealers can short bonds, their gross positions may be more reflective
of their willingness to make markets than their net positions, since market-makers may be keeping
net positions low to limit their exposure to changes in asset prices. This measure points to a less
pronounced decline in dealer activity in US Treasury markets. By comparison, market-making in
Korean debt markets – as gauged from the same metric – would appear to have declined in recent
years, contrasting with the increase in dealer inventories. For India, in turn, this measure seems to
convey the same trends as implied by developments in net dealer positions (Graph 3, right-hand
panel).
200 36 0.4 12
100 24 0.3 9
0 12 0.2 6
–100 0 0.1 3
1 2 3
Includes all US primary dealers. Sample of 10 primary dealers and banks. Domestic central government bonds.
9
These data are based on the Japanese Flow of Funds and ECB statistics. For Japan, “financial
institutions” refers to banks and securities companies, whereas it encompasses all monetary
financial institutions (excluding the Eurosystem) in the euro area.
10
Technically, ETFs redeem creation units only to authorised participants, such as dealers who make
markets for ETF shares in the secondary market, and typically by payment in kind. Hence, ETF
investors rely on the immediacy services provided by these dealers and may need to bear the risk
that ETFs trade at a discount to the fund’s net asset value. Ramaswamy (2011) provides a discussion
of the risks associated with large investor withdrawals.
Rising demand for immediacy services – the case of corporate bonds Graph 4
Bond mutual fund and ETF flows1 Trading volume and turnover ratio2 Bid-ask spreads
USD bn Percent Percent of yield Percent of par value
9 80 1.2 3.0
6 60 1.0 2.5
3 40 0.8 2.0
0 20 0.6 1.5
–3 0 0.4 1.0
Sources: Federal Reserve Bank of New York; FINRA; IMF GFSR; Bloomberg; EPFR; Study Group member contributions based on national
data; BIS calculations.
11
Dick-Nielson (2013), for example, argues that the reduction in US dealers’ corporate bond
inventories is likely to have raised transaction costs in particular for those bonds that dealers cannot
turn over quickly, such as lower-rated issues that are not included in major bond indices.
12
These spreads, which are based on transaction data, may overestimate the degree of market
liquidity. Based on an alternative measure using changes in bond prices, Bao et al (2011) conclude
that US corporate bonds, for example, are far less liquid than implied by their bid-ask spreads.
Box 1
Market resilience can be characterised along several dimensions. One is based on a dynamic definition, suggesting
that market resilience is the time it takes for prices to return to fundamental values in response to a shock (eg an
order imbalance in response to new information). Another uses a functional definition, where market resilience is the
ability of the market to continue functioning in situations of market stress. This is the case if market-makers and
other market participants are able to absorb shocks without a major impact on the immediacy they provide to
buyers and sellers.
A fragile market, by comparison, is one where prices tend to significantly diverge from fundamental values in
response to shocks, with the adjustment often being subject to discontinuous pricing (“gapping”), bouts of volatility
as well as an abrupt deterioration in liquidity conditions. A market is also considered to be fragile if it is often hit by
disturbances particularly in the absence of obvious fundamental drivers.
Factors supporting market resilience include (i) a large and diverse number of active market participants,
(ii) sufficient capacity and willingness of market-makers to absorb short-term order imbalances, (iii) market
transparency, supporting participants in evaluating fundamental asset values as well as assessing counterparty and
other risks and (iv) effective supervision.
On a daily basis, indicators of fragility measure how far market liquidity deviates
from its average over the day and, in particular, how often unusually illiquid
conditions arise. At lower frequencies, these indicators assess how often highly
unusual episodes of illiquidity arise, such as the low depth and high price impact
observed in the US Treasury market on 15 October 2014 (Box 2).
Jumps as indicators of fragile liquidity. One approach to gauging the
fragility of liquidity conditions is to measure the frequency of illiquidity spikes,
defined as an abrupt deterioration (“jump”) in the underlying liquidity metric. The
identification of such jumps is ideally based on high-frequency data which reveal
more about the fragility of liquidity conditions than metrics based on averaging
over a longer time span. In addition, the analysis of complementary liquidity metrics
(eg bid-ask spreads, quoted depth) is likely to provide a more complete picture of
liquidity conditions. Such data are only available for a few highly liquid markets such
as those for major benchmark sovereign bonds or related futures. Yet, any identified
trends in the robustness of liquidity conditions for these markets are likely to have
broader implications, given their role as reference markets for the pricing in other
fixed income securities and related derivatives.
Trading activity surges and liquidity deteriorates. During the event window, trading was continuous. PTFs
and bank-dealers, in that order, accounted for the largest shares of trading volume in both the cash and futures
markets. No trades on the interdealer cash or futures platforms were broken or adjusted, nor was there price
“gapping”, ie jumps from one price to another with no transactions in between. The high trading volume and
continuity of pricing showed that the ability to transact remained in place even at the most volatile times of the day.
Yet, market participants reported significant liquidity concerns and some participants temporarily disengaged their
automated price-quoting systems, turning back to manual or voice trading to manage their risks.
The grey shaded area in the left-hand and centre panels indicates the event window.
1
A passive order is defined as a standing order to buy or sell an instrument in the order book, while an aggressive order is that which is
executed when matched against a standing “passive” order. 2 Inverted scale for the bid-ask spreads quoted by banks and dealers.
PTFs reduce depth while dealers widen spreads. Both PTFs and bank-dealers took action to contain their risk
exposure to volatility during the event window (Graph A). PTFs continued to provide the majority of order book
depth and maintained a tight spread between bid and ask prices, but decisively cut back their limit order quantities.
In contrast, bank-dealers widened their bid-ask spreads so that limit orders were only met at a substantial distance
from the top of the book. Despite the surge in trading volume during the event window, there was no noticeable
change in net positions of PTFs or bank-dealers. This is in contrast to the build-up in short positions in the futures
markets (ie positions that gain in value if the price of the underlying asset declines) by hedge funds during the first
half of the event window (when prices were rising).
Order imbalances. During the event window, an imbalance between the volume of buyer-initiated trades and
seller-initiated trades was observed, with more buyer-initiated trades as prices rose, and more seller-initiated trades
as prices fell. Bank-dealers and PTFs were the main net aggressive buyers of Treasury futures as prices rose, and net
Liquidity provision. A similar breakdown of the net passive trade flow by participant type shows that PTFs
were large net passive sellers during the first part of the event window and net passive buyers during the second
part of the event, consistent with market-making activity and their contribution to the depth at the top of the order
book. Notably, the PTF pattern of passive flows closely mirrors the pattern of PTF aggressive flows. Hence, as a
group, their net position remained largely unchanged throughout the event window, suggesting that the PTFs were
deploying many different types of trading strategy. In contrast, net passive bank-dealer flows were not indicative of
significant market-making activity during the event window.
Self-trading increased. A notable aspect of trading on 15 October was the heightened level of self-trading,
defined as a transaction in which the same entity takes both sides of the trade so that no change in beneficial
ownership results. During the event window, the share of overall transactions resulting from self-trading was
substantially higher than average. Another aspect of self-trading flows during the event window was its directional
nature. Between 09:33 and 09:39, the cumulative net aggressive buyer- minus seller-initiated self-trade volume
increased by around $160 million in the cash 10-year note, accounting for close to one fifth of the total imbalance
between buyer- and seller-initiated trades observed over that time interval. The bulk of self-trading in cash and
futures markets occurred among PTFs, perhaps because such firms can run multiple distinct trading algorithms
simultaneously.
The analysis also revealed that changes to the Treasury market structure over recent years have been
significant. These changes provide an important context for understanding the unusual volatility that day and for
assessing the risk that such an event might recur.
For a more detailed analysis, see Joint Staff Report, “The U.S. Treasury Market on October 15, 2014”, 13 July 2015, on which this case
study draws.
The first row of Graph 5 plots average bid-ask spreads for 10-year US Treasury
notes, JGB futures and Italian Treasury bonds (Buoni del Tesoro Poliennali; “BTPs”),
respectively. The dots indicate instances where spreads have shown day-to-day
jumps (ie increases) of two standard deviations or more. The second row, for
comparison, highlights jumps (ie abrupt declines) in quoted depth for the same
financial instruments.
13
As argued in, for example, Brunnermeier (2009), concerns about banks’ as well as other major
market-making institutions’ default and liquidity risks may have propagated illiquidity risks during
the global financial crisis.
Jumps in bid-ask spreads and quoted depth: signs of fragile liquidity? Graph 5
Spreads: US Treasuries1 Spreads: JGB futures2 Spreads: BTPs3 Jumps over past 250 days
US cents JPY cents Bps Number of jumps
8 4 140 24
6 3 105 18
4 2 70 12
2 1 35 6
0 0 0 0
2009 2011 2013 2015 2009 2011 2013 2015 2009 2011 2013 2015 2009 2011 2013 2015
Spread Jump UST JGB BTP
1 2 3
Depth: US Treasuries Depth: JGB futures Depth: BTPs Jumps over past 250 days
USD m JPY 100 m EUR m Number of jumps
200 400 80 8
150 300 60 6
100 200 40 4
50 100 20 2
0 0 0 0
2009 2011 2013 2015 2009 2011 2013 2015 2009 2011 2013 2015 2009 2011 2013 2015
Quoted depth Jump UST JGB BTP
1
10-year US Treasury notes. 2 Japanese Government Bond (JGB) futures contract based on 10-year JGBs. 3 10-year Italian government
bonds (Buoni del Tesoro Poliennali; “BTPs”). 4 Total number of jumps recorded over the past 250 trading days. A jump is defined as a rise
(decline) in spreads (the percentage change in quoted depth at the first tier) of at least twice its standard deviation.
14
Recent analysis of flash events in futures markets (ie instances where prices swing by unusually
large amounts within a trading hour) – which are typically accompanied by strained liquidity
conditions – suggests that such events may actually be a relatively common feature in markets that
are highly liquid on normal trading days (Massad (2015)).
Between January 2014 and April 2015, yields on German government bond (“bunds”) declined markedly, before
reversing sharply in mid-April and surging again in early June (Graph B, left-hand panel). The rise in yields was
widespread and marked by bouts in volatility, resembling the bond market sell-off during mid-2013 (“taper tantrum”)
that followed a reassessment by market participants of the pace of monetary policy exit in the United States.
Yields and intraday volatility Bid-ask spreads Order book depth and price impact1
Per cent Per cent Bps Bps Bps EUR m
1.8 45 45 6.0 3 24
1.2 30 30 4.5 2 21
0.6 15 15 3.0 1 18
0.0 0 0 1.5 0 15
Q3 14 Q4 14 Q1 15 Q2 15 Q3 14 Q4 14 Q1 15 Q2 15 Jan 15 Mar 15 May 15
Lhs: yields Rhs: intraday volatility Rhs: 10-year Lhs: Rhs:
1
10-year 10-year Lhs: 30-year Bund futures 10-year
30-year 30-year
The first vertical line in each panel shows 7 May 2015; the second one shows 3 June 2015, the date of the ECB policy announcement.
1
The price impact is based on the Eurex liquidity measures, which approximated the roundtrip costs of buying and, at the same time,
selling futures with a notional amount of €50 million.
Sources: Eurex; MTS Euro Benchmark Markets; Thomson Reuters; BIS calculations.
Prices swing in absence of a trigger. On 7 May 2015, price swings within the day were particularly pronounced,
with yields on long-term bunds rising by more than 20 basis points, before returning to their opening prices on the day.
In contrast to the rise in yields in early June, which has been attributed to changes in inflation expectations following the
release of the ECB’s economic outlook (BIS (2015b)), there are no clear indications of any specific events triggering the
market reaction on 7 May. In this regard, the event looks similar to the US market “flash rally” (Box 2).
Market uncertainty. In the absence of a specific trigger, market commentary suggests that a number of factors
may have contributed to comparatively fragile market conditions in the run-up to the yield reversal. One factor is
uncertainty regarding the implementation of the Eurosystem’s Public Sector Purchase Programme (PSPP), including its
impact on the availability of securities for trading that the ECB and national central banks were expected to purchase on
a large scale. Such concerns may have been further fuelled by expectations of a reduction in net issuance of bunds. In
addition, they may have contributed to the perceived decline in market liquidity ahead of the yield reversal, with bid-ask
spreads for long-term bunds trending upwards amid declining depth (Graph B, centre and right-hand panels).
Stretched positioning. An unwinding of positions by leveraged directional investors in fixed income derivatives
markets represents another candidate factor. Reportedly, positions based on expectations that the PSPP would induce a
further decline in interest rates had become relatively crowded during April. This suggests that some participants’
positions may have been vulnerable to even small increases in yields, particularly if faced with margin calls on their
derivatives contracts. Adding to this are indications of less activity in German government bond futures markets in
recent years, as gauged from some measures (eg trading volume relative to open interest). Furthermore, the price
impact of large orders in bund futures contracts spiked in May (Graph B, right-hand panel), underscoring the close ties
between liquidity conditions in cash and futures markets.
3. Drivers
This section discusses structural and conjunctural factors that have a bearing on the
trends presented in Section 2. The structural factors relate to changes in the market
environment due to innovations in technology and to regulation. Their impact on
the way fixed income markets function has been substantial. At the same time,
trends in market liquidity are also driven by powerful conjunctural factors such as
post-crisis deleveraging and the conduct of monetary policy in major currency
areas. The discussion in this section is largely of a qualitative nature given the
difficulties in quantifying the relative influence of these factors on the availability
and pricing of liquidity services.
15
See Markets Committee (2016) for a detailed analysis.
16
See, for example, IMF (2015) or Bessembinder et al (2006).
17
According to PWC (2015), the average transaction size on NYSE and Euronext was about
US$ 10,000 and US$ 13,000 in the year 2015, respectively, By comparison, average trade sizes for
US investment grade corporate bonds and European corporate bonds stood at about US$ 500,000
and US$ 600,000.
The broader post-crisis response represents another key driver of current trends.
Banks, in many jurisdictions, have continued to cut back their trading-related
exposures and have raised their capital buffers (Graph 6, left-hand and centre
panels). Some have reportedly also narrowed the scope of their market-making
activities by focusing on a selection of fixed income instruments or by increased
client tiering (ie offering certain immediacy services only to core clients). In addition,
many market-makers are reportedly providing liquidity only when they can easily
match client orders, but step back from quoting during more volatile market
conditions, particularly in the absence of formal market-making arrangements.
Reassessing the risks. The available evidence suggests that this development
reflects a mix of both market-based and regulatory factors. In terms of market-
based adjustments, diminished dealer risk tolerance as well as a more granular
assessment of risk-adjusted returns for individual business lines have, in many cases,
contributed to reducing the dealers’ willingness to make markets.19 More stringent
regulatory requirements to contain systemic risks in the financial system, in turn,
have scaled down dealers’ risk-taking capacity.
Structural shift or cyclical adjustment? It is hard to predict how far bank
deleveraging reflects a structural trend or a conjunctural development. Research
points to the fact that dealers adjust their leverage procyclically (Adrian and Shin
(2010), (2014)), suggesting that cycles in liquidity conditions are a recurring feature
of financial markets (CGFS (2001)). Likewise, dealers’ shareholders appear to have
18
Some electronic platform providers have developed specific protocols to facilitate the trading of
large amounts, such as “workups” in US Treasury markets. See Fleming and Nguyen (2015) and
Fleming et al (2015).
19
See, for example Adrian et al (2013) and Adrian et al (2015).
Value-at-risk (VaR) leverage1 Dealer VaR in emerging markets Bond valuations and duration2
Per cent Per cent INR bn USD trn Years
8
6.5
6 8 0.3
7
6.0
4 6 0.2 6
5.5
5
2 4 0.1
5.0
4
0 2 0.0 3 4.5
02 04 06 08 10 12 14 16 2011 2013 2015 2007 2009 2011 2013 2015
3
US banks Mexico (lhs): India, avg VaR (rhs): Lhs: Rhs:
Continental European banks
4 Trading Govt securities Market cap Effective duration
6
UK banks
5 VaR Corporate bonds
1
Annualised total trading VaR (99% confidence) divided by total equities, weighted by banks’ total assets. 2 Based on the Merrill Lynch
global corporate bond index. 3 Bank of America, Citigroup, Goldman Sachs, JPMorgan Chase, Lehman Brothers (to Q2 2008), Morgan
Stanley. 4 BNP Paribas, Deutsche Bank, Société Générale, UBS. 5 Barclays, Royal Bank of Scotland, HSBC. 6 In percent of banks’ net
capital.
Sources: Bloomberg; Merrill Lynch; Study Group member contributions based on national data.
20
See Roengpitya et al (2014).
21
While the two, highly simplified, pricing examples (Box 4) do not attempt to assess the quantitative
impact of new regulations on banks’ existing portfolios, the survey results appear broadly in line
with some industry studies. For one example, McKinsey (2011) estimates the return on equity will
decline from 19% (flow rates) and 18% (flow credit) to 8% and 6% due to the impact of recent
regulatory reforms (ie Basel 2.5, Basel III and the OTC derivatives reform). The return on equity for
banks’ proprietary trading, in turn, is expected to decline from 35% to 7%.
22
For examples of central bank surveys that gather information on market liquidity, see the ECB’s
Survey on credit terms and conditions in euro-denominated securities financing and OTC derivatives
markets (September 2015), and the Federal Reserve Bank of New York’s Federal Reserve senior credit
officer opinion survey on dealer financing terms (June 2015). These surveys suggest that market
participants consider regulation to have had a strong negative impact on fixed income market
liquidity.
23
Primary dealer (PD) schemes are one possible example, with PDs supporting liquidity in sovereign
bond markets in exchange for eg preferential access to primary market issuance.
Box 4
Informal survey. To inform the assessment of drivers affecting the supply of immediacy services, the Study Group
conducted an informal survey among market-makers from August to September 2015, receiving responses from
more than 40 firms from 11 jurisdictions.
Survey participants were asked to estimate the costs associated with holding two stylised trading books: (i) a
repo-funded matched domestic sovereign bond book (ie with no directional exposure) and (ii) a net long corporate
bond book, composed of long-term investment grade corporate bonds with interest rate risks being fully hedged.
For both examples, respondents provided estimates of current regulatory capital requirements, those given under
the former Basel II rules and for a fully phased-in Basel III framework. Participants also estimated operational and
trading costs as well as the costs associated with regulatory compliance (eg expenses on reporting, staff).
Scope of the survey. The two examples are based on highly simplifying assumptions and asked respondents
to make additional assumptions if needed for the estimation of specific cost factors, acknowledging differences in
cost accounting across firms, pending finalisation of the Basel III calibration, and the limited time available to
conduct the survey. Therefore, the results provide a broad gauge of the relative importance of these costs, rather
than intending to quantify the specific impact of any individual regulation.
Ex 1: matched sovereign book Ex 2: long corporate book Estimated trading book turnover
Percent of gross revenue Percent of gross revenue Book turnover per year
100 100 50
80 80 40
60 60 30
40 40 20
20 20 10
0 0 0
Basel II Current Basel III Basel II Current Basel III ’10 ’15 ’10 ’15
Sovereign Corporate
Trading and funding costs
Operational costs (excl regulation) 8% return on (Tier 1) capital p a Median 25th 75th
Operational costs (due to regulation) Profit in excess of 8% p a
1
Average estimates based on 42 survey responses. Gross revenues are kept unchanged in all three scenarios and are determined by
assuming a return on total capital of 8% (ie capital charge on required Tier 1 capital) under a fully phased-in Basel III framework. For the
current and fully phased-in Basel III estimates (Basel II), all other costs (ie cost of trading, operational cost) are based on respondents’
estimates for the year 2015 (2010).
Sovereign bonds. Keeping these caveats in mind, the survey responses suggest that, for the sovereign bond pricing
example, the assessment of the main drivers of regulatory capital costs depends on the bank’s current balance
sheet position – some respondents considered the leverage ratio to have the largest impact, whereas others pointed
Corporate bonds. For the corporate bond example, respondents indicated that, on average, revisions to the
Basel II market risk framework (“Basel 2.5”) had had the largest impact on regulatory charges. In line with this,
respondents suggested, on average, that capital charges would have had increased significantly for this pricing
example, when moving from Basel II to current requirements. The remaining phase-in of the Basel III requirements,
in turn, was expected to have only a minor impact. Assuming constant revenues and a return on capital of 8%
annually under the fully phased-in Basel III framework, survey responses suggest that for this example the return on
capital would have amounted to about 26% annually under Basel II requirements (Graph C, centre panel).
Business model adjustments. The majority of respondents suggested that inventory turnover had increased
from 2010 to 2015, particularly for sovereign bonds (Graph C, right-hand panel). This result tallies with the perceived
changes in business models underway, suggesting that many market-makers have trimmed their inventory but are
seeking to maintain their trading activities by raising inventory velocity.
Monetary policy
24
Funding conditions are loosened both directly, given the reduction in policy rates, as well as more
indirectly, given the increase in the value of collateral instruments.
25
For a formalisation of this effect, see eg Gromb and Vayanos (2002) as well as Brunnermeier and
Pedersen (2009). The analysis by, for example, Fontaine et al (2015) provides empirical support for
this channel.
26
According to the IMF (2015), the Federal Reserve’s purchases of MBS initially supported market
liquidity in these instruments, whereas liquidity decreased modestly during the later stages of the
purchasing programme(s) due to scarcity effects.
27
Another factor is increasing issuance by lower-rated companies and in smaller amounts, indicating
that investors are holding an increasing amount of less liquid bonds. See IMF (2015).
Market implications
28
PWC (2015), for example, document a continued decline in the number of active market-makers in
European corporate bonds, particularly for high-yield instruments.
29
How far liquidity providers have acted as shock absorbers in the past is, however, an open question.
Anand and Venkataraman (forthcoming), for example, suggest that liquidity providers without
explicit market-making obligations pull back in unison if market conditions are unfavourable.
30
This risk is not new, however, as suggested by the discussion in Barth et al (2002).
31
Feedback loops in market liquidity may also arise across different assets, as suggested by the
empirical work of Chordia et al (2005) or Goyenko and Ukhov (2009). To explain such effects, Cespa
and Foucault (2014), for example, propose a model in which liquidity providers in one asset infer
information by observing the prices of other securities. If market liquidity in the reference securities
declines, the prices of these securities become less informative. Higher uncertainty, in turn, induces
liquidity providers to cut-back their immediacy services in the market they are active in.
32
Barclay and Henderschott (2004), for example, provide empirical evidence of liquidity externalities
by comparing market conditions during and outside stock exchange trading hours.
33
See, for example, Brunnermeier (2009).
34
This could include policies to ensure that new technologies help preserve a competitive market
structure, while also incentivising participation by heterogeneous market participants (ie in terms of
investment horizons, liquidity needs and risk assessments). See also Markets Committee (2016) for
additional policy considerations that focus on electronic trading.
35
See, for example, recent proposals on best practices for market participants when promoting
liquidity and market robustness (Treasury Market Practices Group (2015)). Another example is the
US Securities and Exchange Commission’s proposals to improve the liquidity risk management
practices of open-ended funds and the permission for “swing pricing” under certain circumstances
(SEC (2015)).
The Group is also grateful to José María Vidal Pastor and Jhuvesh Sobrun (Bank for
International Settlements) for able research assistance.