WPS (DEPR): 01 / 2024

RBI WORKING PAPER SERIES

Pricing of Interdealer Derivatives

in a Limit Order Market

Vidya Kamate
and
Abhishek Kumar

1
DEPARTMENT OF ECONOMIC AND POLICY RESEARCH
MARCH 2024
The Reserve Bank of India (RBI) introduced the RBI Working Papers series in
March 2011. These papers present research in progress of the staff members of
the RBI and at times also those of external co-authors when the research is jointly
undertaken. They are disseminated to elicit comments and further debate. The
views expressed in these papers are those of the authors and not necessarily
those of the institution(s) to which they belong. Comments and observations may
please be forwarded to the authors. Citation and use of such papers should take
into account its provisional character.

Copyright: Reserve Bank of India 2024

 Pricing of Interdealer OTC Derivatives in a Limit Order Market

Vidya Kamate and Abhishek Kumar1

Abstract

Using regulatory interdealer trade-level data on Overnight Indexed Swaps (OIS)

in India, the paper examines the trading behaviour and prices in an interdealer
market populated by traders of varying liquidity needs. Inactive investors, proxied
by traders’ lower level of trading volume in the previous quarter, received a lower
return relative to active investors when trading outside a Central Limit Order Book
(CLOB) venue market but received relatively better returns on it. This differential
return could be attributed to the differing levels of speed preference across
investors with active investors being the most impatient or needing the quickest
execution. Consistent with extant theoretical literature on speed frictions, active
investors preferred trading on CLOB (faster venues). Inactive traders submitted
limit orders and faced slower execution whereas active traders submitted market
orders and received quicker execution on CLOB. The findings allude to the
greater role played by liquidity requirements in a CLOB as opposed to other non-
CLOB OTC markets where search and bargaining frictions dominate.

JEL Classification: G13, G14, G15

Keywords: Overnight indexed swaps, OTC derivatives, limit order book, active
investors, market order

1
Vidya Kamate (vkamate@rbi.org.in) and Abhishek Kumar (abhishekvkumar@rbi.org.in) are Managers from
Department of Economic and Policy Research and Financial Markets Regulation Department, respectively, of the
Reserve Bank of India. They are grateful to Mayank Gupta, DEPR Study Circle participants and an anonymous
external reviewer for comments and suggestions. The views expressed are those of the authors and not of the
Reserve Bank of India.

1
 Pricing of Interdealer OTC Derivatives in a Limit Order Market

Introduction

The size of global derivatives market stands at $598 trillion based on notional
outstanding as on December 20212. A better understanding of these markets is
essential for policymakers and market participants as these markets are large,
opaque, complex and fragmented resulting in significant price heterogeneity.
Consequently, these markets have been a focus of major post-crisis regulatory
reforms including the 2009 Group of Twenty (G20) leaders’ agreement to reform and
strengthen the global financial system. Dealers play a central role in price formation
and liquidity provision in these markets, supplying liquidity to clients in Dealer-to-Client
(D2C) segment and trading among themselves in Dealer-to-Dealer (D2D) or
interdealer segment. While significant recent research has focused on understanding
frictions that drive the price heterogeneity in D2C segment e.g., search and bargaining
(Duffie et al., 2005, Hau et al., 2019), network centrality (Li and Schürhoff, 2019),
inventory holding costs (Colliard et al., 2021) and informational rents (Bolton et al.,
2016), the trading in D2D segment has been assumed to be low cost and frictionless
(Duffie et al., 2005; Cenedese et al., 2020).

Using interdealer trade-level data on Interest Rate Swaps (IRS) in India, we

analyse the factors that determine pricing in interdealer Over-the-Counter (OTC)
derivatives market. By classifying investors into active and inactive categories based
on their trading volume in the previous quarter, we examine whether differential
liquidity needs or levels of impatience across investors has an impact on their trading
behaviour and quality of trade execution. In the setting of a dynamic limit order market
populated by traders of varying level of impatience, Foucault et al. (2005) highlight the
impact of this non-informational friction on market quality and resilience.

In this spirit, we aim to provide evidence of an important and hitherto unexplored

friction viz., differential impatience or need for execution speed that can determine
pricing in D2D derivatives market. We find that inactive investors received lower
returns as compared to active investors when trading outside of a central limit order
book (CLOB)3 venue but received systematically better returns while trading through
it. In other words, traders’ differential liquidity needs affect trading prices, but
magnitude and direction of this pricing impact depends on the trading venue. By

2
BIS Statistics Explorer: Table D5.1 accessed on June 11, 2022 from https://stats.bis.org/statx/srs/table/d5.1?f=pdf
3
A central limit order book is an exchange-style execution method that matches all bids and offers according to price
and time priority. Users can also see order book depth in real time.

2
highlighting the traders’ impatience channel, our paper contributes to the limited
literature analysing frictions in the D2D markets4.

To suitably assess the contribution of this paper, it becomes important to place

in context the way electronification of OTC trading is progressing. There have emerged
a variety of venues featuring a diverse set of execution protocols – CLOBs, Request-
for-Quotes (RFQs), Request-for-Stream (RFS) etc. OTC derivatives are also
increasingly being traded on electronic venues albeit primarily of the nature of RFQ/
RFS etc. The non-standardised nature of such instruments may limit the feasibility of
a CLOB dealing mode. The market for OTC derivatives, like other OTC markets, is
characterised by a two-tier market structure – D2D and D2C segments. In some cases,
the D2D segment, which is primarily a market for risk management, may trade
standardised instruments, thereby, supporting a CLOB venue. In such a D2D setting,
a CLOB platform was introduced in the Indian IRS market in 2015 which has rapidly
become the dominant venue.

We also analyse the effects of investor impatience on the choice of trading

venue and find that active investors have a larger probability of trading on the CLOB
venue as compared to inactive investors5. Despite receiving poorer returns as
compared to inactive investors, active investors prefer to trade on the CLOB venue
due to the accrual of speed-sensitive gains from trades through faster venues.

In CLOB, the general understanding is that participants carry their trades by

submitting either limit orders or market orders. Limit orders are stored in a limit order
book, waiting for future execution (“Limit Day Orders”). Their execution is triggered by
incoming market orders, which are matched with the best offers in the book (Foucault,
1999). Probably to avoid the risk of being “picked off”, a variant of limit order, the
Immediate-or-Cancel (IOC) limit order is executed immediately while cancelling any
unfilled portion. We observe that on the OIS CLOB venue, orders are placed only as
limit orders, with limit day order and limit IOC orders each accounting for about half of
the total trades. Thus, execution generally happens when a limit day order is matched
with either a new limit day order or an IOC order. This indicates that participants exhibit
the tendency to use market orders with limit-IOC order options. In the rest of this paper,
market orders, thus, refer to limit-IOC orders. OIS return, calculated as the order rate
paid/received by the counterparty over the prevailing OIS prices, for trades confirms

4
In this regard, Neklyudov (2019) models dealers differentially in terms of their search intensities and
consequently, different reservation values in equilibrium. Similarly, inventory trading frictions of interdealer
markets have been highlighted in extant literature e.g., Hansch et al. (1998) and Reish and Werner (1998, 2005)
for equity markets; Schultz (2017) for corporate bonds and Paddrick and Tompaidis (2019) for credit derivative
markets.
5
This is in line with the theoretical evidence in Pagnotta and Philippon (2018) which showed that faster venues
attract more speed-sensitive investors.

3
the theoretically-held view that limit orders are executed at better prices than market
orders (in this case, the IOC orders).

We also study how dealers’ impatience affects order placement strategies and
time taken for execution in limit order trading. In agreement with empirical predictions
of Foucault et al. (2005), we find that more active investors have a larger probability
of trading via market orders as opposed to limit orders. Dealers bear waiting costs that
are directly proportional to the time elapsed between order placement and transaction
completion. Therefore, they face a trade-off between execution price and the time
taken for execution. As a result, impatient investors prefer to place market orders over
limit orders6. Relatedly, we also find that the time taken for execution for limit orders
placed by active investors is significantly lower than that of the inactive investors
highlighting the preference for speed among more active investors.

The rest of the paper is organised as follows. Section II provides an overview

of literature, followed by important research questions and testable hypotheses in
Section III. Sections IV and V provide important institutional details of OIS market in
India, and details of the data used in the paper, respectively. Section VI discusses the
main empirical findings of the paper followed by additional robustness tests in Section
VII. Section VIII provides the concluding observations.

II. Literature Review

Our paper, as noted earlier, contributes to the literature on interdealer trading

venue choice and directly relates to the literature that analyses the role of speed-
related advantages in investors’ choice of the trading venue (Pagnotta and Philippon,
2018; and Foucault et al., 2016). Another strand of literature analyses the effect of
transparency and anonymity and predicts that anonymous trading platforms will invite
more informed trades (Röell, 1990; Fishman and Longstaff, 1992, Forster and George,
1992). The empirical evidence on the role of transparency and anonymity in venue
choice has been mixed (Barclay et al., 2003; and Reiss and Werner, 2005). Much of
the literature has focused on the investors’ choice of trading on an OTC vis-à-vis an
exchange market. Lee and Wang (2018), Glode and Opp (2020) and Holden et al.
(2021) developed a model of adverse selection to explain the prevalence of OTC
trading in the presence of exchanges. The data on Indian OIS interdealer market
shows that almost all of the interdealer trades moved to the CLOB trading system after
its introduction indicating potentially lower role of informational frictions perhaps due
to the institutional nature of the market where a majority of the participants are
sophisticated investors.

6
The findings in Keim and Madhavan (1995) support this result. They find that index and technical traders tend
to place market orders (speed preference) whereas value traders tend to place limit orders (price preference).

4
 Our results also relate to the large body of literature on limit order books and
investors’ order placement strategies. The theoretical static models of optimal bidding
strategies are typically based on asymmetric information (Glosten, 1994; Chakravarty
and Holden, 1995; Handa and Schwartz, 1996; Rock, 1996; and Seppi, 1997). Among
the dynamic models, the results in our paper are closest to the predictions generated
by Foucault et al. (2005) and Roşu (2009) that assume the lack of asymmetric
information across traders and waiting costs as the main frictions determining the
choice between limit and market orders.

Our study contributes to the scant but growing literature related to

understanding the microstructure of interest rate derivatives markets. Cenedese et al.
(2020) uses European Trade Repository (TR) data to highlight the role of counterparty
credit risk in OTC derivative pricing. Benos et al. (2020) analysed the role of
centralised trading in determining market quality and liquidity in EUR swap markets.
Given the data availability restrictions, there is very limited research on price
heterogeneity in IRS markets across the globe.

The IRS markets in India provide an ideal setup to analyse the interdealer
liquidity frictions due to the following reasons. First, derivatives trading on a limit order
book market is a unique feature of the Indian market. Generally, derivatives in most
markets are traded through RFQ platforms, if traded electronically at all. Globally, the
market infrastructure has evolved in a manner wherein provisioning of trading and
clearing services are segregated and identifying the trade counterparty is a necessity
for which RFQ venues are more suited. In India, however, a tight integration between
trading, clearing and settlement services has provided the necessary conditions for
provisioning certain services, such as anonymous CLOB venue.

Second, almost all IRS contracts in India (whether executed on OTC non CLOB
or CLOB system) are settled via a CCP which eliminates the counterparty risk friction
in pricing7. Third, informational frictions are less likely to prevail in OIS markets since
investors in OIS D2D markets are a relatively small homogenous group of fairly
sophisticated agents and all the public information relevant for trading gains in OIS
markets arrives at a pre-determined time and uniformly across all agents. Therefore,
there is limited scope for trading based on private information signals.

III. Testable Hypotheses and Research Questions

In this section, we highlight the main hypotheses about the trading behaviour
and prices in interdealer OIS market. The first hypothesis is derived from the

7
On account of home regulations, certain foreign banks operating in India are not permitted to clear trades
through CCIL.

5
theoretical literature on frictions driving pricing differentials in derivatives market. Most
of the extant literature assumes frictionless trading in the D2D segment (Duffie et al.,
2005) and focuses its attention on search, bargaining and/or informational frictions in
the D2C segment. Given this, it becomes pertinent to analyse whether these frictions
also play a role in the D2D segment of OIS markets, or the trading remains frictionless.

Hypothesis 1: Trading in D2D OIS market is not frictionless. In other words, there are
systematic differences in trading execution and prices between certain groups of
investors (active vs. inactive dealers).

A large majority of derivatives, the world over, are OTCs that are voice-traded
bilaterally. Recently, hybrid mechanisms such as RFQ platforms that allow investors
to solicit quotes from multiple dealers simultaneously have begun to be used more
widely. In addition, very few markets like India, have introduced limit order book trading
in D2D OIS market. While OTC markets, due to their non-anonymous nature are
characterised by search and bargaining frictions (Li and Schürhoff, 2019; and Hau et
al., 2021), exchange markets are predominated by speed and informational frictions.
Therefore, a related question that arises is whether the frictions that drive differential
pricing in D2D OIS market, outside of CLOB venue, differ from those that drive pricing
on the CLOB venue? In other words, do price differentials between active and inactive
investors depend on the trading venue? This leads us to the following hypothesis.

Hypothesis 2: Pricing differential between active and inactive investors depends on

the trading venue.

The structure of securities trading has transformed dramatically in the past few
decades with newer venues leading to fragmentation of trading in many markets. In
addition, rapid technological developments have resulted in increased trading speeds
across many instruments, particularly in standardised derivative markets. However, a
significant portion of trading still relies on human inputs. As a result, there is significant
heterogeneity in trading across venues and markets. Extant literature (Pagnotta and
Philippon, 2018) has highlighted fragmentation based on technological improvements
and trading speed with faster venues attracting speed-sensitive investors resulting in
the following testable hypothesis.

Hypothesis 3: Active investors prefer trading on a faster venue.

Traders value order execution speed differently. Traders are likely to choose
different order placement strategies depending on their level of impatience (Foucault
et al., 2005) and hence, their speed preference. Any continuous limit order trading
system comprises limit and market orders. While market orders guarantee immediate
execution at the best available price, limit orders ensure a desired price (or better) but

6
do not guarantee execution. Therefore, traders face a trade-off between waiting costs
and speed premium charges which leads to the following hypothesis.

Hypothesis 4: Inactive traders are more likely to post limit orders and active traders
are more likely to post market orders.

Given the speed preference of active investors, it is probable that out of the
total limit orders placed by all investors, the time for execution of limit orders placed
by active investors is lower than that of the inactive investors. Since the costs of waiting
are larger for active investors, they are more likely to place limit orders with relaxed
limits so that order execution gets a priority over execution price. Therefore, the
following hypothesis arises as a natural corollary.

Corollary to Hypothesis 4: The time taken for execution of limit orders posted by
inactive traders is likely to be higher than the one posted by active traders.

IV. Institutional Details of the Indian OIS Market

Interest Rate Derivatives (IRDs) have been permitted in India since 1999. They
are traded both on OTC, in the form of Interest Rate Swaps (IRS), Forward Rate
Agreements (FRAs), swaptions etc., and on exchanges, primarily as Interest Rate
Futures (IRFs). The IRS are, by far, the dominant IRDs in India. The market is split
into three segments: market-maker only, interdealer or D2D segment, and market
where clients trade with dealers, i.e., the D2C segment. The D2D segment is about
three to four times larger than the D2C segment (in terms of amount outstanding) with
major participation from foreign banks and primary dealers (PDs), followed by private
sector banks. The share of public sector banks, while increasing, remains relatively
small.

The most popular IRS in India, the MIBOR8 OIS is an instrument where the
floating leg of the swap is linked to an overnight index, compounded daily over the
payment period. The instrument is traded on expectations of the counterparties about
the future path of interest rate. OIS contracts are also used as a hedging tool for
investors in government securities. They allow financial institutions to manage various
features of their debt portfolios, including portfolio duration. The open interest in
MIBOR OIS has increased over the years (Chart 1).

8
The overnight Mumbai Interbank Outright Rate (MIBOR), published daily by Financial Benchmark India Pvt.
Ltd (FBIL), is the benchmark rate for call money transactions. It is calculated based on the call money transactions
executed on the NDS-call platform.

7
 Chart 1: MIBOR OIS Trades Outstanding in D2D Market
60 120

50 100
₹ lakh crore

Thousands
40 80

30 60

20 40

10 20

- -
2008-09

2009-10

2010-11

2011-12

2012-13

2013-14

2014-15

2015-16

2016-17

2017-18

2018-19

2019-20

2020-21

2021-22

2022-23 (Upto
August 2022)
Notional Sum Trades, RHS
Note: Trades outstanding at the end of each financial year in MIBOR OIS D2D segment. CCIL
undertakes regular portfolio compression exercises for D2D MIBOR OIS outstanding trades.
Source: CCIL; Authors’ calculations.

OTC derivatives are also increasingly being traded on electronic venues albeit
primarily of the nature of RFQ/RFS etc. The non-standardised nature of such
instruments may limit the feasibility of a CLOB dealing mode. Notwithstanding, in some
cases the D2D segment, which is also a market for risk management for market
makers, may trade standardised instruments, thereby, supporting a CLOB venue. In
such a D2D setting, a CLOB platform was introduced in the Indian IRS market in 2015
which has rapidly become the dominant venue with a share of 65 per cent of overall
market turnover in 2020. Trading volume, which was completely dominated by OTC
non CLOB market in 2014 shifted significantly to the CLOB segment by 2021 (Chart
2).

Chart 2: Evolution of Trading Volume in OIS D2D Market in India

18

15

12
₹ lakh crore

-
Qtr2
Qtr3
Qtr4
Qtr1
Qtr2
Qtr3
Qtr4
Qtr1
Qtr2
Qtr3
Qtr4
Qtr1
Qtr2
Qtr3
Qtr4
Qtr1
Qtr2
Qtr3
Qtr4
Qtr1
Qtr2
Qtr3
Qtr4
Qtr1
Qtr2
Qtr3
Qtr4
Qtr1
Qtr2
Qtr3
Qtr4

2014 2015 2016 2017 2018 2019 2020 2021

Non-CLOB CLOB
Note: The chart shows the quarterly trading volume in OIS D2D markets in India from 2014 to
2021.
Source: CCIL; Authors’ calculations.

8
 Clearing Corporation of India Limited maintains the Trade Repository (CCIL-
TR), where all OTC derivative trades in the Indian market are reported. Most of the
D2D trades are centrally cleared (Chart 3).

Chart 3: Share of Centrally Cleared Trades

100
90
80
70
Per cent

60
50
40
30
20
10
-
Jun-14

Jun-15

Jun-16

Jun-17

Jun-18

Jun-19

Jun-20

Jun-21
Mar-15

Mar-16

Mar-17

Mar-18

Mar-19

Mar-20

Mar-21
Sep-14
Dec-14

Sep-15
Dec-15

Sep-16
Dec-16

Sep-17
Dec-17

Sep-18
Dec-18

Sep-19
Dec-19

Sep-20
Dec-20

Sep-21
Dec-21
Source: CCIL; Authors’ Calculations.

On OIS CLOB venue, the orders are placed only in the form of limit orders, with
limit day order and limit IOC orders each accounting for about half of the total trades.
Thus, execution generally happens when a limit day order is matched with either a
new limit day order or an IOC order. While participants have the option of submitting
a market order, the observations allude towards a tendency to replace market orders
with IOC order option. OIS return, calculated as the order rate paid/received by the
counterparty over the prevailing OIS prices, for trades confirms the theoretically held
view that limit orders are executed at better prices than market orders (or in this case
IOC orders). Relatively limited participant base (only institutional investors), execution
in lot sizes and limited liquidity could be few factors on account of which participants
may prefer to use limit IOC mode of dealing over placement of market orders.

V. Data and Summary Statistics

Transaction-level data for rupee-denominated MIBOR-OIS contracts was

retrieved from CCIL-TR. Each transaction level data point provided information related
to both counterparties (through name); trade date and time; contracted swap rate,
settlement date, notional value of the contract and whether the trade was executed
over CLOB or outside it. We supplement transaction information with order information
for trades executed over CLOB. We study transactions contracted from April 1, 2014
to December 31, 2021, which captures the market transition to the CLOB. To remove
any false or inaccurate reports we only keep trades with a fixed rate that is within 100
basis points (bps) from the benchmark (same maturity) end-of-day swap rate mid-
quote from Bloomberg. In our sample, 54 dealers reported over three lakh transactions
amongst themselves in the D2D segment.

9
 Summary statistics of important variables used in the analysis are presented in
Table 1. A total of 315,128 combined trades were conducted in the D2D market over
the sample period out of which roughly 41 per cent trades were conducted over the
CLOB venue. Return to buyer, defined as the difference between the previous day
closing benchmark rate and transaction level swap rate (fixed leg of the contract)9 is
larger on the OTC non-CLOB segment as opposed to the CLOB segment. Return to
seller is symmetric to the return to buyer variable and defined as the difference in bps
between transaction level swap rate and previous day closing benchmark rate. The
average trade size is roughly 1.4 times larger on the OTC non-CLOB segment at ₹ 89
crore as opposed to the CLOB segment (₹ 62 crore) with the standard deviation being
twice that on the CLOB venue.

Table 1: Trade-Level Summary Statistics

p10 p25 p50 p75 p90 Mean Std. Dev N
Panel A: All Trades
Return to Buyer (4.25) (2) 0.25 2.25 5 0.3 4.5 315,128
Return to Seller (5) (2.25) (0.25) 2 4.25 (0.3) 4.5 315,128
Notional 25 25 50 100 150 77.9 140.7 315,128
Tenor 0.5 1.0 2.0 5.0 5.0 2.7 1.8 315,128
Panel B: Trades on OTC non CLOB
Return to Buyer (4.0) (1.5) 0.5 2.5 5.0 0.4 4.6 186,845
Return to Seller (5.0) (2.5) (0.5) 1.5 4.0 (0.4) 4.6 186,845
Notional 25 25 50 100 200 89 172 186,845
Tenor 0.5 1.0 2.0 5.0 5.0 2.7 1.8 186,845
Panel C: Trades on CLOB
Return to Buyer (4.5) (2.0) 0 2.0 4.5 0.1 4.3 128,283
Return to Seller (4.5) (2.0) 0 2.0 4.5 (0.1) 4.3 128,283
Notional 25 25 45 55 100 62 73 128,283
Tenor 0.5 1.0 2.0 5.0 5.0 2.8 1.9 128,283
Note: Panel A provides summary statistics for the full sample. Panel B and Panel C provide
the summary statistics for trades executed on OTC non CLOB and CLOB, respectively. Return
to Buyer (of interest rate protection) is defined as the difference in bps between the previous
day closing benchmark rate and transaction level swap rate (fixed leg of the contract).
Similarly, Return to Seller is defined as the difference in bps between transaction level swap
rate and previous day closing benchmark rate. Notional refers to the notional value of the swap
contract in INR. Tenor refers to the tenor of the swap contract in years. The numbers within
brackets indicate negative values.
Source: CCIL; Authors’ calculations.

The average trade size over the years across both trading platforms is
highlighted in Chart 4. As can be seen from the plot, the average trade size on the
CLOB system picked up post-2015 after its widespread adoption by the dealer banks
but average trade size for non-CLOB trades remained higher than CLOB for all the
years. This is in line with findings in Holden et al. (2021) that indicate that OTC

9
The return definition is consistent with the ones used in extant literature. See, for example, Cenedese et al.
(2020).

10
segment attracts larger trades due to the price discount the investors are able to
achieve on account of greater bargaining power.

Chart 4: Average Trade Size across Trading Platforms

120

100

80
₹ crore

60

40

20

-
2014 2015 2016 2017 2018 2019 2020 2021
Non CLOB CLOB
Note: Trading on CLOB platform was introduced in 2015.
Source: CCIL; Authors’ calculations.

We classify dealers into active or inactive groups based on their level of trading
volume. A dealer is defined as active if she was in the top quartile of dealers based on
gross notional value traded in the preceding calendar quarter of the trade date and
inactive, otherwise.

Table 2: Summary Statistics for Active Vs. Inactive Dealers

Active Inactive
Variables
All Non CLOB CLOB All Non CLOB CLOB
Buyer
Total Number of Trades 242,574 141,573 101,001 66,790 39,508 27,282
Average Trade Size (₹ crore) 82 92 66 64 76 47
Average Return 0.24 0.43 (0.02) 0.37 0.38 0.34
Difference in Average Return
(6.05) 2.01 (12.12)
of Active vs. Inactive (t-stat)
Seller
Total Number of Trades 245,572 143,601 101,971 63,792 37,480 26,312
Average Trade Size (₹ crore) 81 91 66 66 79 47
Average Return (0.30) (0.39) (0.16) (0.16) (0.52) 0.35
Difference in Average Return
(6.36) 4.68 (16.39)
of Active vs. Inactive (t-stat)
Note: Active dealers represent the top quartile of dealers based on gross notional value traded
in the preceding calendar quarter of the trade date. The remaining dealers are classified as
inactive. Return to Buyer (of interest rate protection) is defined as the difference in bps
between the previous day closing benchmark rate and transaction level swap rate (fixed leg
of the contract). Similarly, Return to Seller is defined as the difference in bps between
transaction level swap rate and previous day closing benchmark rate. The numbers within
brackets indicate negative values.
Source: CCIL; Authors’ calculations.

11
 Table 2 highlights trade level summary statistics by active and inactive dealer
group. As expected, active dealers engage in much larger number of trades, almost
thrice the total number of trades than inactive dealers on CLOB venue as well as
outside of it. The average trade size is significantly larger for active investors as
compared to inactive ones across trading venues, alluding to them being larger and
more sophisticated investors. Active investors receive higher average returns (0.43)
than inactive investors (0.38) when trading outside of CLOB venue but receive lower
average returns (-0.02) than inactive investors (0.34) when trading on them,
highlighting the differential impact across trading venues for active and inactive
dealers. We will explore this differential impact in greater detail in the subsequent
section.

VI. Empirical Results and Discussion

As discussed earlier, the trading in D2D market may not be frictionless. The
frictions affecting D2D pricing may be similar to or different from the ones governing
prices in D2C segment. In line with Hypothesis 1, we test whether similar investor
groups receive differential price execution for the same security on the same day.
Research on corporate bond trading has highlighted execution quality differences
between small and large trades, between frequently and infrequently traded issues,
and also between different customer groups (e.g., Goldstein and Hotchkiss, 2007;
Edwards et al., 2007 and Hendershott and Madhavan, 2015 among others).
Differential prices have also arisen as a result of difference in investor sophistication
in OTC FX markets (Hau et al., 2021). We run trade-level regressions of returns on
contract, trade and dealer-specific characteristics. The results are reported in Table 3.
The first two columns are for trades executed OTC, while columns (3) to (6) show
trades executed on the CLOB venue. The results indicate that active investors receive
systematically better returns as compared to inactive investors over OTC market with
return to seller being systematically higher for active investors as compared to the
inactive ones. The regressions control for notional value of the contract, tenor, date-
specific and counterparty dealer-specific fixed effects. There is no need to control for
credit ratings of counterparties as trades are cleared through a central counterparty
(CCP).

Active investors receive larger returns to the order of 0.575 basis points which
is statistically significant. This result is consistent with the findings in O’Hara et al.
(2018) who found that active insurance companies received better returns on OTC
platforms as compared to inactive ones.

In contrast, the results in columns (3) and (4) suggest that active dealers
received lower returns over the CLOB venue as compared to inactive dealers. Dealers

12
received systematically better returns when they placed limit orders as opposed to
market orders. This is on expected lines given that limit orders, while not guaranteeing
a fast or certain execution, provide a better control on the execution price received by
the dealer. These specifications control for counterparty dealer fixed effects
highlighting that the price differential between active and inactive investors is not being
driven by certain dealer specific factors such as larger inventory costs etc. for a certain
group of dealers. The regressions also control for date fixed effects to account for the
impact of macroeconomic events, such as policy rate announcements that may
influence the returns or volatility in the OIS market on any given day.

In specifications (5) and (6), we introduce an interaction term between active

investor dummy and market order dummy to check if active investors receive
systematically different returns when they place market orders as compared to limit
orders. We find that, while placing market orders results in poorer returns as compared
to limit orders, active investors receive slightly better returns on market orders than
inactive investors.

We identify the underlying explanations for active investors receiving better

returns relative to inactive investors in an OTC market but poorer returns in the CLOB
market. Extant literature has suggested that due to the non-anonymised nature of OTC
markets, search and bargaining frictions are likely to result in better returns for well-
connected and sophisticated investors. O’Hara et al. (2018) found that active
insurance companies received better returns in OTC US corporate bond market due
to dealers’ bargaining power who discriminate against inactive or passive traders. This
could be a potential explanation for active dealers receiving better terms-of-trade as
compared to inactive dealers in the OIS market. This benefit of non-anonymity
disappears in a CLOB market where the identities of trading parties are not known
before a trade is executed.

Next, we attempt to understand the causes of poorer returns to active investors,

vis-à-vis inactive investors in the CLOB market. Foucault et al. (2005) modelled a limit
order market with liquidity traders of varied impatience levels. They argued that traders
value order execution speed differently with more impatient traders facing larger costs
of waiting. Active investors demand liquidity from the market and are likely to be more
impatient since their costs of waiting for trade execution are likely to be larger than
inactive investors. This may cause them to receive poorer returns on the CLOB venue
as greater speed of execution is likely to be preferred over improved execution costs.

13
Table 3: Regression Results of Active versus Inactive Investors’ Returns
over Trading Venues
OTC non-CLOB CLOB
(1) (2) (3) (4) (5) (6)
Variables
Return Return Return Return Return Return
to Seller to Buyer to Seller to Buyer to Seller to Buyer
0.575** -0.153 -0.096 -0.161* -0.221* -0.435***
Active dealer dummy
(0.233) (0.169) (0.107) (0.095) (0.122) (0.114)
0.442*** 0.459***
Limit order dummy
(0.068) (0.076)
-0.713*** -0.999***
Market order dummy
(0.130) (0.153)
Active dealer dummy x 0.336** 0.680***
Market order dummy (0.158) (0.173)
*** **
0.109 -0.190 -0.320 0.286 -0.320*** 0.285**
Log(tenor)
(0.147) (0.177) (0.121) (0.118) (0.121) (0.118)
Trade size large x 0.201 -0.248 -0.059 0.017 -0.059 0.024
Log(notional) (0.685) (0.719) (0.064) (0.067) (0.064) (0.067)
Trade size med x 0.180 -0.231 -0.051 0.010 -0.050 0.017
Log(notional) (1.056) (1.093) (0.051) (0.051) (0.051) (0.051)
Trade size small x 0.237 -0.291 0.032 -0.069 0.033 -0.059
Log(notional) (1.278) (1.314) (0.062) (0.064) (0.062) (0.064)
Date FEs Y Y Y Y Y Y
Counterparty Dealer FEs Y Y Y Y Y Y
Observations 181,262 181,262 128,283 128,283 128,283 128,283
Adjusted R2 0.122 0.122 0.343 0.344 0.343 0.344
Note: Columns 1 to 2 report results of trades executed OTC non-CLOB and columns 3 to 6
report results of trades executed over the CLOB venue. Here and in subsequent tables, Return
to Buyer (of interest rate protection) is defined as the difference in bps between the previous
day closing benchmark rate and transaction level swap rate (fixed leg of the contract).
Similarly, Return to Seller is defined as the difference in bps between transaction level swap
rate and previous day closing benchmark rate. Active dealer dummy takes a value of one if
the dealer of the trade was in the top quartile of dealers based on gross notional value traded
in the preceding calendar quarter of the trade date and zero, otherwise. Limit order dummy
takes a value of one if the trade order is a limit order and zero, otherwise. Market order dummy
takes a value of one if the trade order is an immediate-or-cancel order and zero, otherwise.
Log(tenor) refers to the natural logarithm of tenor of the swap contract in years. Log(notional)
refers to the natural logarithm of the notional value of the swap contract in ₹ crore. Trade size
large is a dummy variable that takes a value of one if the trade size is greater than ₹100 crore
and zero otherwise. Trade size med is a dummy variable that takes a value of one if the trade
size is greater than ₹25 crore and less than equal to ₹100 crore and zero otherwise. Trade
size small is a dummy variable that takes a value of one if the trade size is lesser than or equal
to ₹ 25 crore and zero otherwise. *, **, *** denote statistical significance at 10 per cent, 5 per
cent, and 1 per cent levels, respectively. Standard errors clustered at the trade date level are
reported in parentheses.
Source: CCIL; Authors’ calculations.

We have conjectured that active investors receive poorer returns on the CLOB
venue due to their impatience or speed preference. Next, we test whether their trading
behaviour and order placement strategies are consistent with this line of supposition.
Extant theoretical literature on trading speed and fragmentation (Pagnotta and

14
Philippon, 2018) suggests that faster venues attract speed-sensitive investors. This
suggests that if active dealers are indeed impatient investors, then they should prefer
to trade over faster venues i.e., CLOB venue over OTC non-CLOB market which
brings us to the test of Hypothesis 3. We test whether active dealers are more likely
to trade on the CLOB venue vis-à-vis OTC non-CLOB. Table 4 presents results of
trade-level regressions where the dependent variable takes a value of one if the trade
was conducted over the CLOB venue and zero, otherwise10. The main independent
variable of interest is active dealer dummy, the coefficient of which is positive and
statistically significant highlighting that active investors are more likely to trade on the
CLOB trading system. This preference for trading over the CLOB venue of active
dealers is consistent with their preference for speed of trade execution and faster
venues.
Table 4: Regression Results of the Choice of Trading Venue for Dealers
CLOB Venue Dummy
Variables (1) (2)
Sell Side Buy Side
0.010** 0.010***
Active dealer dummy
(0.004) (0.004)
-0.045*** -0.039***
Log(tenor)
(0.003) (0.003)
-0.064*** -0.062***
Trade size large x Log(notional)
(0.003) (0.003)
-0.080*** -0.077***
Trade size med x Log(notional)
(0.004) (0.004)
-0.080*** -0.075***
Trade size small x Log(notional)
(0.004) (0.004)
Date FEs Y Y
Counterparty Dealer FEs Y Y
Observations 300,911 300,911
Adjusted R2 0.458 0.460
Note: The table reports trade-level OLS regression results of CLOB Venue dummy on dealer
and trade characteristics. Column (1) reports regression results from the perspective of the
seller and column (2) reports regression results from the perspective of the buyer. The
dependent variable takes the value one if a trade is executed over the CLOB and zero,
otherwise. Other variables have the same meaning as described in Table 3. *, **, *** denote
statistical significance at 10 per cent, 5 per cent, and 1 per cent levels, respectively. Standard
errors clustered at the trade date level are reported in parentheses.
Source: CCIL; Authors’ calculations.

In addition to a preference for faster venues, extant theoretical literature on

order placement strategies indicates that impatient investors prefer to place market
orders over limit orders in an LOB setting. Market orders represent the demand for

10
The results reported in Tables 4 and 5 are based on an ordinary least squares (OLS) regression. The exercise
was also repeated using the probit model; the results from the OLS and probit model-based regressions were found
to be similar.

15
immediacy of execution whereas limit orders represent supply of immediacy to other
dealers. A limit order is able to improve upon the cost of execution of an order at the
expense of the speed of trading. In line with Hypothesis 4, if active investors represent
impatient investors, then they should prefer to place market orders over limit orders.

Table 5 presents results of trade-level regressions of the choice of limit versus

market orders on dealer and trade characteristics. The dependent variable is a limit
order dummy that takes a value of one if the executed trade is a limit order and zero,
if it was a market order. The coefficient of active investor dummy is negative and
statistically significant implying that consistent with their speed preference, active
investors have a lower probability of placing a limit order vis-à-vis inactive investors.
In other words, active investors prefer to trade via market orders on the LOB.
Interestingly, larger trades have a higher probability of being placed via limit orders
since the benefits of improved execution price with a limit order are likely to be higher
with larger trade sizes.

Table 5: Regression Results of the Choice of Order Type

of Trades on CLOB Venue
Limit Order Dummy
Variables (1) (2)
Sell Side Buy Side
-0.115*** -0.089***
Active dealer dummy
(0.007) (0.006)
-0.007* 0.013***
Log(tenor)
(0.004) (0.004)
0.0001 0.027***
Trade size large x Log(notional)
(0.005) (0.005)
-0.003 0.016***
Trade size med x Log(notional)
(0.006) (0.005)
-0.006 0.011
Trade size small x Log(notional)
(0.007) (0.007)
Date FEs Y Y
Counterparty Dealer FEs Y Y
Observations 128,283 128,283
Adjusted R2 0.121 0.125
Note: The table reports results of all trades executed over the CLOB venue. Column (1)
reports OLS regression results from the perspective of the buyer dealer and column (2) reports
OLS regression results from the perspective of seller dealer. The dependent variable takes
the value one if a trade is a limit order trade and zero, otherwise. Other variables have the
same meaning as described in Table 3. *, **, *** denote statistical significance at 10 per cent,
5 per cent, and 1 per cent levels, respectively. Standard errors clustered at the trade date level
are reported in parentheses.
Source: CCIL; Authors’ calculations.

In line with the speed preference of active dealers, as a natural corollary to the
order choice hypothesis, the time to trade execution should be lower for active
investors. This should hold across all order types on the CLOB venue. Table 6

16
presents average time taken for execution in seconds for active and inactive dealers
by order type. We can clearly see that average time taken for execution for active
dealers is lower than inactive dealers and the difference in execution time is larger for
limit order trades with time taken for execution for inactive dealers almost 1.5 times
that of active dealers. This may happen on account of wider limits set by active dealers
as compared to inactive ones, to ensure improved trade execution probability.

Table 6: Time to Execution for by Order Type

Execution Time (in seconds) Limit Order Market Order
Sell Side
Active Investors 1,760 0.1
Inactive Investors 2,474 0.3
Buy Side
Active Investors 1,523 0.5
Inactive Investors 2,295 0.0
Source: CCIL; Authors’ calculations.

Table 7 reports results of regressions of time taken for trade execution on CLOB
trading system as a function of dealer and trade characteristics. Sample in Columns
(1) and (2) covers all trades executed on the CLOB venue whether they are limit orders
or market orders. Results show that active investors receive a quicker execution by
about 0.11 seconds on the selling side and about 0.08 seconds on the buying side as
compared to inactive investors. The difference in trade execution time is primarily
driven by limit order trades where active dealers receive significantly quicker execution
vis-à-vis inactive dealers. There is no significant difference in execution time across
active and inactive dealers for market order trades. This result could be a result of
wider price bands set by active investors on limit orders ensuring a preference of
speed over price.

Table 7: Regression Results of Time to Execution of Trades over CLOB Venue

All CLOB Trades Market Order Trades Limit Order Trades
Variables Sell Side Buy Side Sell Side Buy Side Sell Side Buy Side
(1) (2) (3) (4) (5) (6)
-0.113*** -0.084*** -0.0002 0.001 -0.007** -0.002
Active dealer dummy
(0.006) (0.006) (0.001) (0.001) (0.004) (0.004)
-0.006 0.015*** 0.0001 0.0004 0.002 0.005**
Log(tenor)
(0.004) (0.004) (0.001) (0.001) (0.002) (0.003)
Trade size large x -0.008 0.015*** 0.0003 0.002* -0.010*** -0.013***
Log(notional) (0.005) (0.005) (0.001) (0.001) (0.003) (0.003)
Trade size med x -0.007 0.009 0.001 0.002* -0.005* -0.009***
Log(notional) (0.006) (0.006) (0.001) (0.001) (0.003) (0.003)
Trade size small x -0.008 0.006 0.001 0.003** -0.001 -0.007*
Log(notional) (0.007) (0.007) (0.001) (0.001) (0.004) (0.004)
Date FEs Y Y Y Y Y Y
Counterparty Dealer FEs Y Y Y Y Y Y
Observations 128,283 128,283 55,697 56,664 72,581 71,613
Adjusted R2 0.131 0.132 0.009 0.010 0.127 0.114

17
Note: Columns 1 to 2 report results for all trades executed over the CLOB venue, with columns
3 and 4 covering only market/IOCC order trades and columns 5 and 6 covering only limit order
trades. Other variables have the same meaning as described in Table 3. *, **, *** denote
statistical significance at 10 per cent, 5 per cent, and 1 per cent levels, respectively. Standard
errors clustered at the trade date level are reported in parentheses.
Source: CCIL; Authors’ calculations.

VII. Robustness Checks

VII.1. Alternative Definition of Active and Inactive Dealers based on Number of Trades

The analysis of differential returns across active and inactive dealers relies on
the definition of activity based on the ranking of dealers according to the trading
volume in the past quarter. We check whether the results are robust across different
definitions of active vs inactive dealers. We, alternatively, define active dealers as
those that belong to the top quartile of dealers based on total number of executed
trades in the preceding calendar quarter. Using this alternative definition, we replicate
the regression results of Table 3 and report them in Table 8 below. The results are
quantitatively similar to the results obtained in Table 3 i.e., active dealers received
systematically better returns as compared to inactive dealers over OTC non-CLOB
market with return to buyer being systematically higher for active investors as
compared to the inactive ones and that active dealers received lower returns over the
CLOB trading system as compared to inactive dealers. Dealers received
systematically better returns when they placed limit orders as opposed to market
orders. Specifications (5) and (6) highlight that while placing market orders results in
poorer returns as compared to limit orders, active investors received slightly better
returns than inactive investors on market orders.

Table 8: Regression Results of Active versus Inactive Investors’ Returns

over Trading Venues
OTC (non CLOB) CLOB
(1) (2) (3) (4) (5) (6)
Variables
Return Return Return Return Return Return
to Seller to Buyer to Seller to Buyer to Seller to Buyer
0.501** -0.210 -0.055 -0.140 -0.187 -0.418***
Active dealer dummy
(0.236) (0.164) (0.118) (0.098) (0.129) (0.118)
0.446*** 0.462***
Limit order dummy
(0.068) (0.075)
-0.733*** -1.014***
Market order dummy
(0.141) (0.161)
0.104 -0.192 -0.320*** 0.285** -0.320*** 0.284**
Log(tenor)
(0.146) (0.176) (0.121) (0.118) (0.121) (0.118)
**
Active dealer dummy x 0.352 0.685***
Market order dummy (0.170) (0.180)
Trade size large x 0.200 -0.248 -0.062 0.013 -0.063 0.017
Log(notional) (0.684) (0.719) (0.064) (0.066) (0.064) (0.066)

18
 Trade size med x 0.179 -0.232 -0.054 0.006 -0.054 0.009
Log(notional) (1.056) (1.093) (0.052) (0.051) (0.052) (0.051)
Trade size small x 0.237 -0.292 0.029 -0.073 0.028 -0.069
Log(notional) (1.277) (1.315) (0.062) (0.064) (0.062) (0.063)
Date FEs Y Y Y Y Y Y
Counterparty Dealer
Y Y Y Y Y Y
FEs
Observations 181,262 181,262 128,283 128,283 128,283 128,283
Adjusted R2 0.122 0.122 0.343 0.344 0.343 0.344
Note: Columns 1 to 2 report results of trades executed OTC non-CLOB and columns 3 to 6
report results of trades executed over the CLOB venue. Other variables have the same
meaning as described in Table 3. *, **, *** denote statistical significance at 10 per cent, 5 per
cent, and 1 per cent levels, respectively. Standard errors clustered at the trade date level are
reported in parentheses.
Source: CCIL; Authors’ calculations.

VII.2. Alternative Definition of Active and Inactive Dealers based on top decile of gross
notional

We, alternatively, define active dealers as those that belong to the top decile of
dealers based on total gross notional value traded in the preceding calendar quarter
of the trade date. We replicate the regression results of Table 3 using this alternative
definition and report them in Table 9 below. The results are quantitatively similar to the
results obtained in Table 3 i.e., active dealers received systematically better returns
as compared to inactive dealers in the OTC market with returns to buyer being
systematically higher for active investors as compared to the inactive ones and that
active dealers received lower returns over the CLOB trading system as compared to
inactive dealers.

Table 9: Regression Results of Active versus Inactive Investors’ Returns

over Trading Venues
OTC (non CLOB) CLOB
(1) (2) (3) (4) (5) (6)
Variables
Return Return Return Return Return Return
to Seller to Buyer to Seller to Buyer to Seller to Buyer
Active dealer dummy 0.661* -0.539 -0.123 -0.103 -0.184* -0.339***
(0.359) (0.404) (0.080) (0.066) (0.105) (0.076)
Limit order dummy 0.445*** 0.468***
(0.068) (0.073)
-0.521*** -0.756***
Market order dummy
(0.083) (0.090)
Active dealer dummy x 0.141 0.525***
Market order dummy (0.135) (0.114)
Log(tenor) 0.118 -0.201 -0.320*** 0.284** -0.319*** 0.288**
(0.150) (0.183) (0.121) (0.118) (0.121) (0.118)
Trade size large x 0.195 -0.230 -0.053 0.017 -0.054 0.025
Log(notional) (0.683) (0.711) (0.065) (0.065) (0.065) (0.065)
Trade size med x 0.174 -0.212 -0.046 0.010 -0.046 0.018
Log(notional) (1.055) (1.084) (0.052) (0.050) (0.052) (0.050)

19
Trade size small x 0.224 -0.270 0.038 -0.068 0.036 -0.062
Log(notional) (1.274) (1.305) (0.062) (0.063) (0.063) (0.063)
Date FEs Y Y Y Y Y Y
Counterparty Dealer FEs Y Y Y Y Y Y
Observations 181,262 181,262 128,283 128,283 128,283 128,283
Adjusted R2 0.122 0.122 0.343 0.344 0.343 0.344
Note: Columns 1 to 2 report results of trades executed OTC and columns 3 to 6 report results
of trades executed over the CLOB venue. Other variables have the same meaning as
described in Table 3. *, **, *** denote statistical significance at 10 per cent, 5 per cent, and 1
per cent levels, respectively. Standard errors clustered at the trade date level are reported in
parentheses.
Source: CCIL; Authors’ calculations.

VIII. Conclusion

Interdealer markets play a pivotal role in liquidity provision and price discovery
in OTC derivatives. Veering away from the assumptions of early theoretical models on
OTC market frictions, trading in interdealer markets is not frictionless, leading to
asymmetric trade execution costs and price heterogeneity across agents and over
time. As opposed to D2C markets, such trading frictions have been relatively
understudied in D2D markets, in general, and OIS markets, in particular.

Using novel and hitherto unused trade-level data on Indian interdealer OIS
market, the paper provides empirical evidence of an important factor in interdealer
markets, namely, the differential liquidity needs or levels of impatience of dealers. We
find evidence of active dealers receiving better returns than inactive ones on OTC non
CLOB markets and the opposite result holding true on a CLOB platform. Consistent
with the speed friction channel, we find that active dealers have a higher probability of
placing market orders as opposed to inactive dealers on limit order trading systems.
Also, we find evidence of speed friction influencing venue choice of dealers, with active
dealers more likely to trade on CLOB platform as opposed to inactive dealers.

In addition to providing empirical evidence of the speed friction channel, this

paper contributes to the debate around the benefits or costs associated with different
trading mechanisms i.e., OTC non CLOB vis-à-vis CLOB. Unlike the recent evidence
on interdealer FX market in other economies (Holden et al., 2021), we find that after
introduction of CLOB mechanism, a large part of the trading moved away from the
OTC non-CLOB market to the CLOB one. This result has the potential to inform policy
on parallel introduction of suitable trading platforms in other OTC markets. The
differential return to active investors on the CLOB mechanism alludes to the potential
benefits for venues to offer vertically differentiated products.

20
References
Barclay, M. J., Hendershott, T., & McCormick, D. T. (2003). Competition among
trading venues: Information and trading on electronic communications networks.
The Journal of Finance, 58(6), 2637-2665.

Benos, E., Payne, R., & Vasios, M. (2020). Centralized trading, transparency, and
interest rate swap market liquidity: Evidence from the implementation of the dodd–
frank act. Journal of Financial and Quantitative Analysis, 55(1), 159-192.

Bolton, P., Santos, T., & Scheinkman, J. A. (2016). Cream‐skimming in financial

markets. The Journal of Finance, 71(2), 709-736.

Cenedese, G., Ranaldo, A., & Vasios, M. (2020). OTC premia. Journal of Financial
Economics, 136(1), 86-105.

Chakravarty, S., & Holden, C. W. (1995). An integrated model of market and limit
orders. Journal of Financial Intermediation, 4(3), 213-241.

Colliard, J. E., Foucault, T., & Hoffmann, P. (2021). Inventory Management, Dealers'
Connections, and Prices in Over‐the‐Counter Markets. The Journal of Finance,
76(5), 2199-2247.

Duffie, D., Gârleanu, N., & Pedersen, L. H. (2005). Over‐the‐counter markets.

Econometrica, 73(6), 1815-1847.

Edwards, A. K., Harris, L. E., & Piwowar, M. S. (2007). Corporate bond market
transaction costs and transparency. The Journal of Finance, 62(3), 1421-1451.

Fishman, M. J., & Longstaff, F. A. (1992). Dual trading in futures markets. The Journal
of Finance, 47(2), 643-671.

Forster, M. M., & George, T. J. (1992). Anonymity in securities markets. Journal of

Financial Intermediation, 2(2), 168-206.

Foucault T. (1999). Order flow composition and trading costs in a dynamic limit order
market. Journal of Financial Markets, 2(2), 99-134.

Foucault, T., Kadan, O., & Kandel, E. (2005). Limit order book as a market for liquidity.
The Review of Financial Studies, 18(4), 1171-1217.

Foucault, T., Hombert, J., & Roşu, I. (2016). News trading and speed. The Journal of
Finance, 71(1), 335-382.

Glode, V., & Opp, C. C. (2020). Over-the-counter versus limit-order markets: The role
of traders’ expertise. The Review of Financial Studies, 33(2), 866-915.

Glosten, L. R. (1994). Is the electronic open limit order book inevitable? The Journal
of Finance, 49(4), 1127-1161.

21
Goldstein, M. A., Hotchkiss, E. S., & Sirri, E. R. (2007). Transparency and liquidity: A
controlled experiment on corporate bonds. The Review of Financial Studies, 20(2),
235-273.

Handa, P., & Schwartz, R. A. (1996). Limit order trading. The Journal of Finance, 51(5),
1835-1861.

Hansch, O., Naik, N. Y., & Viswanathan, A. S. (1998). Do inventories matter in

dealership markets? Evidence from the London Stock Exchange. The Journal of
Finance, 53(5), 1623-1656.

Hau, H., Hoffmann, P., Langfield, S., & Timmer, Y. (2021). Discriminatory pricing of
over-the-counter derivatives. Management Science, 67(11), 6660-6677.

Hendershott, T., & Madhavan, A. (2015). Click or call? Auction versus search in the
over‐the‐counter market. The Journal of Finance, 70(1), 419-447.

Holden, C. W., Lu, D., Lugovskyy, V., & Puzzello, D. (2021). What is the impact of
introducing a parallel OTC market? theory and evidence from the chinese interbank
fx market. Journal of Financial Economics, 140(1), 270-291.

Lee, T., & Wang, C. (2018). Why trade over-the-counter? when investors want price
discrimination. When Investors Want Price Discrimination (February 5, 2018).
Jacobs Levy Equity Management Center for Quantitative Financial Research
Paper.

Li, D., & Schürhoff, N. (2019). Dealer networks. The Journal of Finance, 74(1), 91-144.

Loon, Y. C., & Zhong, Z. K. (2016). Does Dodd-Frank affect OTC transaction costs
and liquidity? Evidence from real-time CDS trade reports. Journal of Financial
Economics, 119(3), 645-672.

Keim, D. B., & Madhavan, A. (1995). Anatomy of the trading process empirical
evidence on the behavior of institutional traders. Journal of Financial Economics,
37(3), 371-398.

Neklyudov, A. (2019). Bid-ask spreads and the over-the-counter interdealer markets:

Core and peripheral dealers. Review of Economic Dynamics, 33, 57-84.

O’ Hara, M., Wang, Y., Zhou, X, (2018). The execution quality of corporate bonds.
Journal of Financial Economics. 130 (2), 308-326.

Paddrik, M. E., & Tompaidis, S. (2019). Market-Making Costs and Liquidity: Evidence
from CDS Markets. SSRN Working Paper

Pagnotta, E. S., & Philippon, T. (2018). Competing on speed. Econometrica, 86(3),

1067-1115.

22
Reiss, P. C., & Werner, I. M. (1998). Does risk sharing motivate interdealer trading?
The Journal of Finance, 53(5), 1657-1703.

Reiss, P. C., & Werner, I. M. (2005). Anonymity, adverse selection, and the sorting of
interdealer trades. The Review of Financial Studies, 18(2), 599-636.

Rock, K. (1996). The specialist’s order book and price anomalies. The Review of
Financial Studies, 9, 1-20.

Röell, A. (1990). Dual-capacity trading and the quality of the market. Journal of
Financial Intermediation, 1(2), 105-124.

Roşu, I. (2009). A dynamic model of the limit order book. The Review of Financial
Studies, 22(11), 4601-4641.

Schultz, P. (2017). Inventory management by corporate bond dealers. Available at

SSRN 2966919.

Seppi, D. J. (1997). Liquidity provision with limit orders and a strategic specialist. The
Review of Financial Studies, 10(1), 103-150.

23