Competition, Liquidity and Volatility - A Comparative Study of

Bombay Stock Exchange and National Stock Exchange

Chandrasekhar Krishnamurti
Division of Banking and Finance
Nanyang Business School
Nanyang Technological University
Nanyang Avenue
Singapore 639798
Tel: (65) 790-5702
Fax: (65) 791-3697
E-mail: ackrishna@ntu.edu.sg
Eugene Lim
National University of Singapore
2
Competition, Liquidity and Volatility - A Comparative Study of
Bombay Stock Exchange and National Stock Exchange
Abstract
India currently has two major stock exchanges: The Bombay Stock Exchange and the
National Stock Exchange. There are important differences in ownership structure,
geographic reach, internal control systems and institutionalised risk management
facilities between the Bombay Stock Exchange and the National Stock Exchange.
The purpose of this study is to examine if these significant structural differences
between these stock exchanges contribute to variations in observed measures of
quality of markets. We use a paired comparison approach and document significant
differences in liquidity and price volatility between the two markets.
3
1.0 Introduction
Provision of liquidity is the major function of a stock exchange. Extant
research has focused on various measures of liquidity in order to quantify the quality
of liquidity services. More recently, research attention has focused on comparing the
relative efficacy of various trading systems and stock exchanges
1
. The major intent of
such studies is to highlight the differences in the quality markets and to determine the
basis for the observed variation.
While there has been plethora of studies concerning the quality of stock
exchanges of developed countries, there has been negligible research attention
devoted to the comparative study of stock markets in emerging countries. Such
research serves two useful purposes. First, they help to identify potential efficiencies
of the different markets and facilitates additional investments in them. Second, it
helps discretionary investors to choose the market with the lowest transaction costs.
Technology has also played an important role in how the markets are structured and
how the stock markets can compete between themselves. This paper concerns itself
with recent developments in Indian stock markets, where there are two major stock
exchanges competing between themselves. We try to provide empirical evidence on
measures of quality of these stock exchanges.
There have been important structural changes in the Indian financial sector.
One of them is the stock market reforms. Bombay stock exchange (BSE) used to be
the premier exchange in India until 1994. The National stock exchange (NSE) began
its operations in 1994 and has dramatically transformed the Indian stock markets.
From the inception NSE started stealing a march over BSE both in terms of trading
performance and in establishing itself as the foremost stock exchange in the country.

1
See for instance the work of Huang and Stoll (1996).
4
They pioneered a completely transparent computerised trading network, which covers
the entire country.
NSE has within a short period of time, won popular acclaim. However, it is an
empirical issue to determine if in fact, NSE provides a better quality market compared
to BSE. The primary aim of this paper is to provide a methodology and empirical
evidence regarding this important issue.
The paper is organized as follows. The next section provides an overview of
the stock market reforms in India, which served as a precursor to the establishment of
NSE. In section 3, we provide a brief comparison of BSE and NSE. Section 4,
describes the data and methodology used in the study. The empirical findings are
outlined in section 5. In the final section, we offer our conclusion and discuss the
implications of our research results.
2.0 Stock market reforms in India
Despite a seemingly high savings rate, India, faced a severe shortage of
investible resources. By early 1990s it was recognized that it is crucial to raise funds
from abroad to fill the gap. Financial sector reforms were needed to remedy the
structural weaknesses and inefficiencies in the stock markets, primary markets,
banking and insurance sectors. Reforms were required in order to boost investor
confidence and broaden the investor base. The Indian corporate sector demanded
these reforms in order to reduce the cost of capital and to enhance its competitiveness.
The government seriously deliberated about these reforms in order to facilitate
participation by foreign institutions and corporates.
Prior to the creation of the National Stock Exchange (NSE), India ranked last
among 12 emerging markets on settlement, safekeeping and operational risk. Poor
liquidity and high bid-ask spreads were common. The system was vulnerable to
5
defaults. The following major problems were frequently cited: outdated trading and
settlement mechanisms, lack of transparency in transactions, delays in physical
delivery and registration of transfer, poor disclosure, price manipulation, insider
trading, lack of liquidity and excessive speculation.
The Government of India appointed the Narasimhan Committee to study the
implementation of possible measures to strengthen Indias financial sector. Among
the major measures recommended by the Committee were: free pricing of initial
public offers, and provision of access to Indian firms to tap global capital markets
through global depository receipts, and foreign currency convertible bonds. The other
steps advocated by the Narasimhan committee are: empowerment of the regulator
Securities Exchange Board of India (SEBI); access of foreign institutional investors
(FIIs) and non-resident Indians to Indian capital and stock markets.
The Pherwani Committee which was appointed to specifically look at stock
exchange reforms proposed the following actions: setting up of a National Stock
Market System (NSMS); uniform settlement across exchanges; shortening of the
settlement cycle; electronic transfer of funds and scripless trading; and the
establishment of a central depository trust. It must be noted that the NSMS envisages
an electronic linkage of principal national and regional exchanges.
The Bombay Stock Exchange (BSE), which was the premier stock exchange at
that point in time was not enthusiastic to these proposals. As a result the NSMS did
not take off. The main reason is that BSE is a brokers club. They knew that NSMS
could not take off without their support. BSE continued to exploit its preeminent
position in the Indian stock markets. As might be expected, BSE also took little steps
to put its own house in order. BSE brokers opposed regulators strictures tooth and
6
nail they resorted to lawsuits and strikes. Another development at this juncture was
the mushrooming of illegal stock exchanges all over the country.
Stock exchange reforms were seen as an essential component of the reform
process to attract foreign investment. The government was increasingly frustrated
with BSEs recalcitrant attitude. The Development Financial Institutions were asked
to set up a modern electronic stock exchange drawing upon the recommendations of
the Pherwani Committee. The National Stock Exchange was set up in response to the
request from the government of India to make the Indian stock markets more efficient
and transparent.
The National Stock Exchange started trading in November 1994. NSE introduced
for the first time, a national network of computerized trading, a clearinghouse, and
special facilities for institutional investors and fully automated screen based trading.
NSE is a completely order driven market, while BSE at that point used a system of
market makers. BSE and other regional stock exchanges used floor-trading systems
at that time. NSE also shortened the settlement cycle in its trading.
NSE has transformed the Indian stock markets in several ways. First, it has
brought about increased transparency to the Indian stock market. Second, it has built
up a more efficient settlement and delivery system. Third, it has improved risk
management systems by the institution of collateralization based on the risk of a
traders position. NSE has carried out these reforms through a combination of new
technology, new processes and enforcement of new regulations.
NSE has largely succeeded in implementing many of the reforms suggested by the
Pherwani Committee. It must be noted that BSE had failed to implement them
initially. One of the reasons behind the success of NSE is that it has a completely
different organizational structure compared to BSE. In NSE membership on the
7
exchange is separated from the ownership. A Board of Directors appointed by the
promoters consisting of a group of Development Financial Institutions, Public Sector
Banks and other Financial Institutions manages the exchange. In BSE ownership and
membership of the exchange are not separated. BSE is essentially a brokers club.
The benefits from NSE initiated reforms permeated to all stock exchanges in
India. First, it brought about increased competition. For the first time, BSE really felt
the heat of a serious competition and was forced to shed many of its investor-
unfriendly customs. A second fall-out is the increase in overall trading volumes.
Trading volumes increased due to two reasons. Some investors shifted their trading
from illegal to recognized stock exchanges. Trading also increased as investors
experienced a user-friendlier environment with regard to settlement and delivery
systems. A third outcome was the lower transaction costs bid-ask spreads paid by
investors.
The entry of NSE brought about major repercussions to most participants in
Indias stock markets. The price of a seat on BSE fell from about Rs. 40 million to
Rs. 10 million. Bombay Stock Exchange launched screen based trading system called
BOLT - Bombay On-line Trading System - in response to NSEs online trading
system. Other Regional Exchanges also followed suit in establishing their own online
trading systems. In terms of regulation, NSE leads while others follow suit rather
reluctantly. In SEBIs books, NSE is a good boy whose example others are urged to
follow.
3.0 A comparison of BSE and NSE
3.1 The Stock Exchange, Mumbai (BSE)
Started as the "The Native Share and Stockbrokers Association" in 1875, BSE is
the oldest stock exchange in India. In fact it is the oldest exchange in Asia, beating
8
even Tokyo Stock Exchange (which was founded in 1878) to the honor. For many
years, BSE had been the premier exchange in the country. It had been referred to as
the Gateway to the capital market in India, [and] a linchpin of the Indian Capital
market.
2
As would be expected of a near monopolistic entity, BSEs investors were
pretty much dependent on the stockbrokers for fairness and good service. This
situation however would change as Indian economy opens up. Faced with economic
deregulation and increased local competition, BSE had to keep its operations at par
with international standards, and to make for a more transparent and more liquid
market.
3.2 National Stock Exchange (NSE)
NSE is the newcomer in the financial markets scene in India. The impetus for
its establishment [had] came from policy makers in the country, who had in mind to
set up facilities [that would] serve as a model for the securities industry in terms of
trading systems, practices and procedures.
3
The exchange was founded as recently as 1994. The NSE Wholesale Debt
Market started its operations on June 30, 1994 while the NSE Capital Market segment
started on November 3, 1994. In 1996, NSE set up the Settlement Guarantee Fund,
and launched indices like NIFTY (NSE-50 Index), NIFTY Junior (Midcap-50 Index)
and DEFTY Index (Dollar denominated Nifty Index).
One important differentiation of NSE from most other exchanges in India is
that in NSE, the ownership and management of the exchange is completely separated
from the right to becoming a trading member. A Board of Directors manages the
exchange. The Board reports to an Executive Committee, which includes
representatives from the Trading Members, the public and the management, delegates

2
Information about BSE could be obtained via its website at http://www.bseindia.com
3
Information about NSE could be obtained via its website at http://www.nse-india.com
9
decisions relating to market operations. Furthermore, the exchange operates various
committees to advise it on areas such as good market practices, settlement procedures,
risk containment systems etc. Industry professionals, Trading Members and exchange
staff manage these committees. The day to day management of the exchange is
delegated to the Managing Director who is supported by a team of professional staff.
These institutional features present a breath of fresh air to the general trend in
Indian stock exchanges, whereby membership on an exchange also meant ownership
of the exchange. A separation of the above should prevent clash of interests on the
part of the member/management, when safeguarding the investing publics welfare.
4.0 Data and methodology
4.1 Data
The primary data used for this paper is high-frequency transaction record of all
trades for a sample of 26 issues cross-listed on both exchanges (See Table 1). The
observations were for a period of 42 days, from 1 January 1997 to 6 March 1997. BSE
and NSE on request made the data available.
Table 1 presents 26 paired issues from each exchange. Table 2 shows a
sample of the data set. For constructing the liquidity measure, trading hours for the
exchanges was taken to be 1000 hours to 1600 hours, and any transactions outside this
time range was ignored in the analyses.
4.2 Methodology
The liquidity measure used in this paper is the Market Efficiency Coefficient
(MEC) developed by Hasbrouck and Schwartz (1987). There are a few advantages for
using this measure. Firstly, the primary data does not differentiate between buy and
sell transactions, and MEC does not require this differentiation for its construction.
Secondly, volume of transactions does not play a role in the construction
10
Table 1
Paired samples from NSE and BSE
Company Name Symbol
Asea Brown Boveri Limited Abb
Andhra Valley Power Supply Company Limited Andravally
Arvind Mills Limited Arvindmill
Ashok Leyland Limited Ashokleyland
Asian Paint India Limited Asianpaint
Bharat Heavy Electricals Limited Bhel
Castrol India Limited Castrol
Cochin Refineries Limited Cochinref.
Colgate Palmolive India Limited Colgate
East India Hotels Limited Eihotel
Grasim Industries Limited Grasim
Housing Development Finance Limited Hdfc
Hindustan Lever Limited Hindlever
Industrial Credit and Investment Corporation of India Limited Icici
Industrial Finance Corporation of India Limited Ifci
Indian Hotel Company Limited Indhotel
Indo Gulf Fertilizers and Chemicals Corporation Limited Indogulf
Indian Rayon and Industries Limited Indrayon
Mahindra & Mahindra Limited M&M
Mangalore Refinery and Petrochemicals Limited Mrpl
Nestle India Limited Nestle
Ponds India Limited Ponds
Ranbaxy Laboratories Limited Ranbaxy
Tata Chemicals Limited Tatachem
Tata Tea Tatatea
Thermax Limited Thermax
Table 2
Sample of Primary Data
Issue Date Time Exchange No. of shares traded Price
EIHOTEL 97-02-27 10:10 BSE 50 410
EIHOTEL 97-02-27 10:21 BSE 150 410
EIHOTEL 97-02-27 10:28 BSE 50 413.75
EIHOTEL 97-02-27 10:43 BSE 50 413.5
EIHOTEL 97-02-27 10:47 BSE 50 408
EIHOTEL 97-02-27 11:09 BSE 50 407.75
EIHOTEL 97-02-27 11:10 BSE 50 407.75
EIHOTEL 97-02-27 11:10 BSE 50 408
EIHOTEL 97-02-27 11:20 BSE 50 407.75
EIHOTEL 97-02-27 11:36 BSE 50 407.75
of this measure, and thus one does not have to worry about the misleading
effect that volume has on conventional liquidity ratio. Thirdly, this measure takes into
consideration the need to distinguish between liquidity and efficiency in the
marketplace, by factoring in both short-period and long-period price volatilities.
11
However, Schwartz (1988) warns that it may be tempting to believe that liquidity can
be measured by using very short period price movements without separately
accounting for the impact of news. This is not correct; news also affects short period
price behavior.
We describe below the procedure for computing MEC.
An issues price relative over the long period may be expressed as the product of price
relatives over T shorter periods:
P
T
/ P
0
= P
1
/ P
0
X P
2
/ P
1
. P
T
/ P
T-1
.(1)
Taking the logarithms of the above equation gives:
R
L
= R
S,t
Where R
L
is the long period logarithmic return and R
S,t
is the short period
logarithmic returns. For the analysis, long period will be taken to be two days and
short period will be taken as one half-hour, and the trading hours are taken to be 1000
hours to 1600 hours. Thus for every two trading days, there would be one R
L
and
twenty-four R
S,t
s computed.
The next step is to find price volatilities over the long and short periods, which
are measured by the variances, Var(R
L
) and Var(R
S,t
) respectively. Hasbrouck and
Schwartz (1987) show that in the absence of execution costs, and assuming
informationally efficient markets, the implied variance of half-hour returns, Var(R
S
)*,
would be given by : Var(R
L
) / 24
MEC is then taken to be the ratio of implied volatility to observed volatility:
MEC = Var(R
S
)* / Var(R
S
) = Var(R
L
) / 24 Var(R
S
)
Execution cost, C, can then be estimated using MEC:
C = [ 0.5 Var(R
S
) ( 1 MEC ) ]
1/2
> 0 for MEC less than 1
12
And
C = [ 0.5 Var(R
S
) ( MEC 1 ) ]
1/2
> 0 for MEC greater than 1
There is one obvious limitation of using MEC to measure liquidity. In the
event that MEC is greater than unity, the estimated C would be negative, which is
meaningless. This can happen when other aspects of the markets operations such as
marketmaker intervention to stabilize trading, the presence of stale limit orders on the
book, sequential information arrival, and inaccurate price determination that involves
partial adjustment to news (Hasbrouck and Schwartz (1987)) dampen short period
volatility.
Hasbrouck and Schwartz (1987) however also posited that factors causing
excessive short period volatility should dominate the factors dampening short period
volatility for most issues, and on average, MECs should be less than unity. (Schwartz
(1988) has found that MECs based on short period variances are predominantly less
than unity) returns.
The empirical analysis has two objectives. Firstly, it aims to establish whether
liquidity level differs across the two markets. Secondly, it aims to uncover some of
the variables that move liquidity in the two markets. By establishing these two
empirical effects, it is hoped that the liquidity-market microstructure relationship in
NSE and BSE can then be better isolated and discussed.
5.0 Empirical Findings
In table 3, we show paired comparisons of the mean and standard
deviation values of liquidity measures and firm characteristics of our sample.
Our results indicate that MEC is consistently higher on the NSE as compared to
BSE. In a frictionlesss market MEC would equal unity. We interpret our results
to mean that trading frictions on NSE is less as compared to BSE. The other
13
important measure, execution cost, C, measured both in percent and in the Indian
currency Rupees, is significantly lower in NSE as compared to BSE. The
average execution cost in NSE is less than half the average value of the cost in
BSE.
We note that trading frequency is higher on the NSE as compared to BSE,
while the average size per trade is higher on the BSE. It would be interesting to
examine if these differences in trading characteristics on the two exchanges have
an effect on MEC and C.
Table 3
Paired Comparison of Liquidity Measures and Firm Characteristics
Variable
Total Sample NSE BSE
Number of
issues
46 23 23
MEC 0.4585 0.6275 0.2896
(0.2932) (0.2443) (0.2376)
C(percent) 0.0065 0.0037 0.0092
(0.0041) (0.0019) (0.0038)
C(Rupees) 2.9220 1.6818 4.1622
AP 5.6279 5.6273 5.6286
(1.0102) (1.0215) (1.0217)
AST 5.3103 5.1459 5.4748
(0.7157) (0.6586) (0.7466)
ANT 4.6252 4.7651 4.4853
(0.9250) (0.8879) (0.9597)
MC 7.5452 7.5452 7.5452
(0.7424) (0.7508) (0.7508)
Note: i) The results are reported for 23 issues in each exchange, omitting Icici, Mrpl and
Ranbaxy, which were found to be outliers, ii) Standard deviations are given in
parentheses.
The Variables are described below:
AP: The logarithm of the average price of an issue in Rupees over the observation period.
AST: The logarithm of the average size per trade of an issue over the observation period.
ANT: The logarithm of the average number of trades per day, for an issue over the
observation period.
MC: The logarithm of the market capitalization for an issue in 10 millions of Rupees.
14
Table 4
Regression with MEC as dependent variable
Equation Constant AP AST ANT MC DNSE R-square
1 -0.9843 0.1063 0.1677 0.0979 -0.0904 0.3658 0.3752
(-1.6297) (1.5488) (2.3840) (1.9119) (-1.1529) (5.0923)
2 -0.4520 0.0923 0.0494 0.0020 0.3544 0.3538
(-0.8950) (1.7876) (1.1990) (0.0379) (4.8774)
3 0.1757 -0.0070 0.0514 -0.0103 0.3235 0.3038
(0.4652) (-0.1338) (1.0289) (-0.1374) (4.4028)
4 -0.5490 0.0261 0.1165 0.0072 0.3763 0.3347
(-0.9511) (0.4652) (1.7357) (0.1164) (5.0910)
5 -0.9648 0.0461 0.1329 0.0595 0.3650 0.3701
(-1.5916) (1.0308) (2.0834) (1.5237) (5.0617)
6 0.1744 0.0549 -0.0174 0.3225 0.3201
(0.4673) (1.3028) (-0.3349) (4.4635)
7 -0.4250 0.0969 0.0244 0.3698 0.3471
(-0.8379) (1.8728) (0.5029) (5.1429)
8 -0.4375 0.0919 0.0500 0.3541 0.3692
(-1.3439) (1.8419) (1.3164) (4.9605)
9 0.2256 -0.0350 0.0346 0.3379 0.3028
(0.6018) (-0.7849) (0.5698) (4.6802)
10 0.1449 -0.0121 0.0474 0.3246 0.3200
(0.4821) (-0.3334) (1.1822) (4.4980)
11 -0.5282 0.0300 0.1185 0.3769 0.3503
(-0.9736) (0.6796) (1.8493) (5.1766)
12 0.4026 -0.0201 0.3379 0.3138
(1.9359) (-0.5606) (4.7177)
13 -0.2130 0.0918 0.3681 0.3584
(-0.7616) (1.8253) (5.1702)
14 0.0656 0.0499 0.3239 0.3341
(0.3607) (1.2796) (4.5376)
15 0.2397 0.0066 0.3379 0.3090
(0.6432) (0.1349) (4.7019)
Note: i) For a 46-issues sample, ii) T-statistics are given in parentheses, iii) R-square is adjusted for
degree of freedom and iv) average R-square is 0.3375. v) DNSE: An intercept dummy variable that is
assigned a value of 1 for all NSE issues and a value of 0 for BSE issues.
15
The purpose of tables 4 and 5 is to find out whether firm-specific trading
characteristics or the stock market used has a greater impact on liquidity and
execution costs. In table 4, we regress MEC, the liquidity measure, on the different
variables outlined above. The price per share and market capitalization do not have
an effect on MEC. The average size per trade has a beneficial impact on MEC. The
number of transactions variable has a weak positive effect on MEC. The most
significant impact comes from the dummy variable, DNSE, which takes the value 1
for NSE and 0 for BSE. The empirical evidence indicates that MEC is significantly
higher on NSE as compared to BSE even after controlling for firm specific variables
that have an effect on trading.
In table 5, we regress C (x 100) on the same variables as before. The dummy
variable, DNSE, is the only one that is statistically significant. The coefficient of this
variable is negative, indicating that firms have significantly lower execution costs on
NSE as compared to BSE. None of the other variables have any discernible effect on
C.
Overall, we can conclude that NSE provides a more liquid market than BSE as
evidenced by lower execution costs and higher MEC. The size per trade exerts a
positive influence on the observed MEC. We also notice that BSE has a higher mean
size per trade compared to NSE. It appears that the NSE system provides a more
liquid market for the stocks trading in it.
The observation that the newer NSE provides better liquidity is interesting.
However, NSE also shows a higher average number of transactions relative to BSE.
Earlier studies have documented that number of trades has a positive significant
impact on price volatility. If that is so, then NSE should have a higher volatility
compared to BSE. An implication of higher volatility on NSE is that it would make it
16
Table 5
Regressions with C ( x 100 ) as dependent variable
Equation Constant AP AST ANT MC DNSE R-square
1 2.3636 -0.1145 -0.1733 -0.0779 0.0668 -0.5865 0.4582
(3.0254) (-1.2900) (-1.9044) (-1.1757) (0.6589) (-6.3125)
2 1.7901 -0.0920 -0.0256 -0.0327 -0.5743 0.4495
(2.7642) (-1.3903) (-0.4851) (-0.4923) (-6.1634)
3 1.1650 0.0025 -0.0299 -0.0160 -0.5428 0.4235
(2.4401) (0.0384) (-0.4726) (-0.1689) (-5.8446)
4 2.0173 -0.0507 -0.1326 -0.0108 -0.5949 0.4532
(2.7750) (-0.7184) (-1.5682) (-0.1392) (-6.3910)
5 2.3492 -0.0700 -0.1476 -0.0495 -0.5860 0.4657
(3.0291) (-1.2235) (-1.8078) (-0.9906) (-6.3507)
6 1.1655 -0.0311 -0.0134 -0.5425 0.4372
(2.4715) (-0.5843) (-0.2038) (-5.9403)
7 1.7761 -0.0944 -0.0443 -0.5822 0.4595
(2.7706) (-1.4438) (-0.7227) (-6.4068)
8 1.5485 -0.0852 -0.0350 -0.5694 0.4594
(3.6988) (-1.3285) (-0.7176) (-6.2021)
9 1.1360 0.0188 -0.0420 -0.5512 0.4342
(2.4218) (0.3369) (-0.5527) (-6.1011)
10 1.1172 -0.0054 -0.0361 -0.5411 0.4369
(2.9395) (-0.1176) (-0.7116) (-5.9308)
11 1.9861 -0.0566 -0.1356 -0.5959 0.4660
(2.9066) (-1.0184) (-1.6798) (-6.4974)
12 0.9211 0.0007 -0.5512 0.4433
(3.5397) (0.0151) (-6.1512)
13 1.3911 -0.0852 -0.5792 0.4655
(3.9232) (-1.3354) (-6.4159)
14 1.0818 -0.0350 -0.5414 0.4498
(4.7084) (-0.7102) (-6.0058)
15 1.1284 -0.0270 -0.5512 0.4459
(2.4336) (-0.4430) (-6.1652)
Note: i) For a 46-issues sample, ii) T-statistics are given in parentheses, iii) R-square is adjusted for
degree of freedom and iv) average R-square is 0.4498.
17
less attractive as a trading venue. We therefore, examine this issue empirically.
For this part of the study, we use the price volatility measure, PV, which is
defined as follows:
PV
it
= ABS [ ( P
it
P
it-1
) / P
it-1
]
where P
it
is the closing price of issue i on day t, and P
it-1
is the closing price of the
same stock on day t-1. PV is calculated for each trading day for each firm.
Table 6
Comparative statistics on volatility, trading frequency and trade size
Variable Total Sample NSE BSE
Number of issues 46 23 23
PV 0.0195 0.0184 0.0206
(0.0248) (0.0202) (0.0286)
N 4.3942 4.5380 4.2504
(1.1356) (1.0931) (1.1593)
AS 5.0436 4.9598 5.1274
(0.9411) (0.8111) (1.0489)
Note: i) For 1886 observations, ii) Standard deviations are given in parentheses
The variables used as described below:
AS: The logarithm of the daily average size of trade for an issue
N: The logarithm of the daily number of trades for an issue.
In table 6, we show mean and standard deviation values for price volatility,
trade size and number of trades for our sample on the two exchanges. We find that
volatility is lower on NSE as compared to BSE. The daily number of transactions is
higher on NSE compared to BSE, but BSE has larger size of trade on average.
Jones, Kaul and Lipson (1994) have shown that the number of transactions has
a positive impact on price volatility and that trade size does not have an effect. In
table 7, we show results of regressing our price volatility measure on trade size and
number of transactions. We confirm the strong effect of number of transactions and a
weak effect of trade size on price volatility for our sample. The dummy variable,
DNSE, has a negative and statistically coefficient. This indicates that price volatility
is lower on NSE as compared to BSE even after controlling for number of trades and
18
trade size. Our empirical evidence indicates that NSE provides a better quality market
for traders on account of its higher liquidity, lower execution costs and lower price
volatility.
Table 7
Regression with PV as dependent variable
Equation Constant N AS DNSE R-Square
1 -0.0038 0.0052 0.0005 -0.0036 0.0572
(-1.0600) (10.4121) (0.7774) (-3.2106)
2 -0.0016 0.0052 -0.0037 0.0574
(-0.7248) (10.6197) (-3.3125)
3 0.0139 0.0013 -0.0020 0.0034
(4.3005) (2.1633) (-1.7286)
Note: i) For 1886 observations, ii) T-statistics are given in parentheses,
iii) R-square is adjusted for degree of freedom and iv) average R-square is 0.0393.
v) DNSE: An intercept dummy variable that is assigned a value of 1 for all NSE issues
and a value of 0 for BSE issues.
6.0 Conclusion
6.1 Structural explanations
By using MEC and price volatility as measures of liquidity, it has been established
that liquidity is indeed better in NSE than in BSE. This is in sync with the expectation
held before empirical evidence made its entry. The last major task left in this paper is
to put in order the consequential differences in the two exchanges market
microstructure, which had initially led to the prima facie expectation.
It has been mentioned that fairness in a marketplace has substantial effects on its
liquidity. NSE is superior in this department on many counts. To begin with, the
impetus for founding NSE had been to buck the trend of slack regulations and to
challenge BSEs near monopolistic hold on Indians capital market scene. The
separation of management from membership in NSE ensures that a trading members
interests shall not override the interests of the exchange as a whole. Such protection
19
is not assured in BSE, where membership is an automatic privilege of ownership and
management.
Another count on fairness from NSE lies in its rules for order matching priority. In
NSE, strict price priority followed by time priority is observed. In the BSE however, a
jobber with a history of large number of transactions can influence the priority of
order matching. As such, investors trading via junior jobbers face possible
sidestepping in their wait to transact. Cohen, Maier, Schwartz and Whitcomb (1986)
have discussed the evidence that violation of priority rule can lead to thinner market
and consequently, illiquidity.
The reputation of the surveillance system in NSE is better too. Although, with
the inception of NSE, BSE had been forced to keep up with market transparency and
regulation, it has only been presenting investors with a minimum acceptable level of
such goods. Coincidentally, at the time of our data analysis, the Securities and
Exchange Board of India had sacked the president of BSE for alleged price
manipulations
4
. With headlines like this one, it is difficult to expect investors to keep
faith in BSEs surveillance system and the receipt of equitable treatment in the
exchange.
The presence of a share depository in NSE also gives it an edge in liquidity.
Without a share depository, the settlement of transactions is much slower as investors
have to wait for administrative work on scrip transfers to be done at the registrar level.
As one cannot sell what (s) he does not own yet, the waiting time taken up for scrip
transfer actually holds up the potential for further transactions. This inevitably slows
down trading volume, which in turn has adverse effect on liquidity. Furthermore, the
time taken up for scrip transfer also affects the rate of information flow in an

4
http://www.economictimes.com/today/25lead01.htm, MAR 25 1999 TOP-STORIES; BSE chief
Parekh gets Sebi sack, HC stay
20
exchange. For example, the lack of identity of share ownership can hide the intention
of a takeover, and this delay of information flow adds to the price volatility when the
takeover intention becomes clear. Such price volatility again adds to illiquidity.
NSE adopts a completely order driven system while BSE has a system that is
part order driven and part quote driven. An order driven system is one where bid-ask
spreads are driven by the agency market-style limit order book. A quote driven system
is like that of a dealer market system, where dealers quote bid-ask spreads.
Furthermore, BSEs choice of a mixed system presents ambiguity to investors. There
is no standard specifying when spread would be order driven, and when it is subjected
to jobbers quote intervention.
Both exchanges have price stabilization features. NSE has price freezes while
BSE has price protection limits. Such features should be arguably having the same
effect on liquidity. The differentiating factor, however, is that NSE is reputed to be
stricter and more partial in its employment of price freezes. BSE on the other hand,
somewhat chooses the instants to trigger the price protection limits, particularly with
regards to the company whose issue is in question. This lack of equitability inevitably
causes investors to question BSEs integrity, which will ultimately affect trade
volume.
6.2 Implications
Before the founding of NSE, BSE had accounted for about 90% of equity trade
volume in India. At last count, BSE and NSE together accounts for about 80% of
volume. There is little reason to believe why NSE should not continue to overtake
BSE.
The only possible exception is the existence of the Badla system in BSE. Such a
derivative-like system is absent in NSE, and could have helped to generate continued
21
interest in BSE. The Badla system allows investors to cut transaction costs by taking
forward position instead of actual position in shares, and the limit order book in BSE
is thicken by such forward positions. As such, liquidity level in BSE is still not as bad
as it should given its structural inferiority. However, one should bear in mind that
derivative instruments could result in speculation-driven price volatility, and the
Badla system is only unique as far as derivative markets are not in place in India.
When such markets are in place, the Badla system would be rendered redundant.
In the light of the empirical evidence, we take the view that BSEs survival is in
jeopardy, unless it takes additional steps to improve the quality of its market. BSE
must of its on accord undertake structural reforms to before more efficient and better
regulated.
22
References
Cohen, K., S. Maier, R. Schwartz, and D. Whitcomb, 1986, The microstructure of
security markets. Prentice-Hall, Englewood Cliffs, NJ.
Hasbrouck, J. and R. Schwartz, 1987, Liquidity and execution costs in equity
markets. Journal of Portfolio Management, Summer, 54-62.
Huang, R. and H. Stoll, 1996, Dealer versus Auction Markets: A paired comparison
of execution costs on NASADAQ and the NYSE. Journal of Financial
Economics, 313-357.
Jones, C.M., G. Kaul, and M.L. Lipson, 1994, Transactions, volume and volatility,
Review of Financial Studies 7, 631-651.
R. Schwartz, 1988, Equity markets Structure, trading and performance. Harper and
Row, New York.
23
Figure 1.
TRADING VOLUMES OF BSE &
NSE
0
5000
10000
15000
20000
25000
A
P
R
_
9
6
J
U
L
Y
_
9
6
O
C
T
_
9
6
J
A
N
_
9
7
A
P
R
_
9
7
J
U
L
Y
_
9
7
O
C
T
_
9
7
J
A
N
_
9
8
T
R
A
D
I
N
G

V
O
L
U
M
E

I
N

R
S
.

M
I
L
L
I
O
N
S
NSE AVG
BSE AVG