Motivation

A

The first aspect of liquidity is a static concept: When an order is placed, by how much does the actual execution price degrade, when compared with a benchmark price? This is accurately measured using “impact cost”, which can be observed exactly on the electronic order matching markets, such as those we have in India.

The second aspect of liquidity is “resilience”. A resilient financial market is one which is able to absorb large shocks and rapidly revert to efficient pricing and high levels of liquidity. The issue of resilience is relevant regardless of the “direction” of the shock: a large upward or a downward movement is equally challenging for market efficiency. When a large order is placed in a market that lacks resilience, it takes a long time for liquidity and price to restore to normal levels. When a market lacks resilience, negative price shocks can adversely affect liquidity. The most extreme version of a lack of resilience is a case where market institutions collapse when there is a large price shock, and trading stops completely.

Resilience is important from two points of view. At an analytical level, resilience is a necessary condition for market efficiency. In an efficient market, prices rapidly revert to fundamentals, regardless of whether the shocks are from the order flow or from news. At a more practical level, poor resilience directly translates to “liquidity risk“. At future dates, economic agents face the risk that they might not be able to execute transactions as required, or the cost of transacting might be prohibitive.

A generic issue which affects these discussions is the problem of measuring instantaneous liquidity. The best measure of instantaneous liquidity is impact cost, which is observed at three time points per day on the National Stock Exchange (NSE). However, impact cost is not readily measureable on the Bombay Stock Exchange (BSE) or on the debt market. Hence, we fall back on other proxies of liquidity: the number of shares traded and the turnover ratio (TR). The number of shares is a better measure than rupee turnover, because it does not get affected by a movement in prices, and it can be summed across exchanges. The TR is defined as the latest one year of rupee turnover divided by the current market capitalisation. This can be measured for both stock exchanges and for the bond market.

In this article, we offer some insights into resilience of the Indian securities markets. We start with the equity market and closely examine the experience of two large price shocks on the equity market

– the 36 per cent drop in the price of Infosys over two days starting April 10, 2003 and the 12 per cent drop in the market index on May 17, 2004. Our findings suggest that the equity market was quite resilient in the face of such shocks.

We go on to compare the impact of price shocks on the liquidity of the equity and debt markets. We find that while the equity market appears to have considerable resilience, this is not the case with the debt market.

Nifty on May 17, 2004: A Natural Experiment

General elections took place in 2004 with an unexpected outcome. In response to the perceived negative news, the Nifty dropped by an unprecedented 21 per cent in the first two hours of trading on May 17, 2004. At the end of the trading day, the market closed with a largest-ever oneday drop of 12 per cent. The CMIE COSPI index, which is a more comprehensive index covering all significant Indian equities, also dropped by 12 per cent.

Despite such a large drop, the equity market institutions did not collapse. Instead, markets stayed open for business.

What is more, the liquidity in the market remained sound. Table 1 presents the evidence of the number of trades that were executed on the NSE and the BSE during the period. We also look at the number of shares that were traded at the time. Both the number of trades and the number of shares traded show no drop in value around May 17, when the index dropped significantly.

Price discovery at this time was concentrated on the index derivatives market (with the largest activity being focused on Nifty futures trading). Here too the liquidity remained strong, as can be seen in the data on number of contracts traded in the last column in Table 1. This experience compares well with (say) the October 1987 crash on the New York Stock Exchange, when liquidity vanished in response to a 21 per cent drop in the index.

Infosys on April 10, 2003: A Natural Experiment

On April 10, 2003, Infosys Technologies, one of the biggest firms in the country, announced a gloomy earnings

The H T Parekh Finance Forum is edited and managed by Errol D’Souza, Shubhashis Gangopadhyay, Subir Gokarn, Ajay Shah and Praveen Mohanty.

Table 2: Price and Turnover of Infosys in April 2003

0 5 10 15 Pre-event Apr 4 1073.80 1119145 Apr 7 1095.72 963174 Apr 8 1058.13 954262 Apr 9 1037.99 979878

outlook. Infosys signalled that it was pre-event days, where the few days Event Apr 10 762.44 4822068

unlikely to be able to maintain the high with event-related elevated turnover were

Apr 11 663.34 6524268

earnings growth associated with the soft-removed, the results were as follows. The Apr 15 714.92 3412717 ware industry. coefficient of before was -0.13315 with a Apr 16 756.13 2586204

Apr 17 740.85 2122830

The valuation of the market dropped t statistic of -1.805. The R2 was 0.04. When

Apr 21 733.15 1261185 sharply and dramatically. Over a two-day the dependent variable was re-expressed Apr 22 729.92 1322230 Post-event

period, the price fell by 36 per cent, one as the log of the number of shares traded,

Apr 23 720.75 1056386

of the largest-ever two-day price changes the coefficient of before was -0.10917 with Apr 24 724.46 1610804 of a large stock in India. However, as a t statistic of -1.622. The R2 was 0.03. Apr 25 727.48 749380

Apr 28 723.21 810668

Table 2 shows, the supply of liquidity on These results suggest that, if anything, the the market remained steady. The number of shares transacted per day, summing

Table 1: Liquidity in the Indian Equity Markets around May 17, 2004

across NSE and BSE, rose sharply on the date of the announcement (April 10) but Date CMIE COSPI Index NSE+BSE Turnover Index Futures (No of Trades) (No of Shares) (No of Contracts)

then fell back to pre-event levels.

Table 2 shows details of the liquidity as Pre-event May 3 741.97 1,649,915 245,354,000 333,921

the news broke on April 10, 2003. The

May 4 755.92 1,531,144 235,415,000 275,337

average level of turnover, prior to the event,

May 5 762.92 1,515,675 241,438,000 231,047was 1,025,670 shares per day. The event May 6 774.88 1,563,617 273,661,000 238,171 was associated with a massive increase in May 7 764.70 1,612,551 284,350,000 253,403

May 10 750.96 1,465,055 242,245,000 278,484

trading, as myriad speculators participated

May 11 717.84 1,381,364 227,157,000 274,407in price discovery. From April 10 to 17, May 12 723.11 1,662,059 247,569,000 292,879turnover increased to the range from two May 13 722.57 2,231,378 363,509,000 484,365 to 6.5 million shares a day. The highly May 14 660.18 2,307,533 389,614,000 497,610

Event

negative news event did not generate a

May 17 582.27negative impact upon liquidity. After this Post-event news was absorbed into the price, the May 18 627.36 1,889,804 294,170,000 272,737 May 19 658.61 2,065,262 307,024,000 306,470

liquidity of the market returned to normal

May 20 653.67 1,840,805 273,963,000 272,847

conditions. The mean turnover in the

May 21 656.47 1,605,056 215,305,000 336,86630 trading days from April 23 onwards was May 24 674.29 1,631,292 220,922,000 309,148 1,169,973 shares per day, which was in-May 25 673.13 1,741,467 229,012,000 365,793

May 26 673.68 1,658,692 224,941,000 333,396

significantly different from the mean turn-

May 27 667.18 1,523,038 204,337,000 353,083

over of the 30 days prior to April 10.

May 28 635.19 1,835,976 259,607,000 294,658In an OLS regression on intercept, May 31 619.22 1,816,211 237,583,000 284,215 with a dummy variable “before” for the

Economic and Political Weekly August 12, 2006

16 14	Figure 2: Bond Market TR 10-year interest rate Bond market turnover ratio (right scale)	200 180 160
12		140
10		120
8		100 80
6		60

10-year zero y-coupon rate

1998 2000 2002 2004

This figure juxtaposes the daily time-series of the 10-year rate on the NSE zero coupon yield curve, against

the monthly time-series of the bond market TR.

India had a remarkable set of years, from 1999 to 2004, where interest rates fell dramatically. This is shown in Figure 1. The bond market TR was 84 per cent in August 1999. It rose dramatically to a level of 194 per cent in April 2002 and stayed at such high levels till September 2003. However, by August 2005, the turnover ratio had dropped back to 84 per cent, back to the levels of August 1999. This suggested a lack of progress on liquidity across a period of six years.

As Figure 2 illustrates, these dates tally up exactly with the story of interest rates. Bond market liquidity was strong when interest rates were falling, i e, when positive price changes were being experienced.

Table 3: Comparing Resilience of theEquity and Bond Markets

Equity		Debt
Coefficient	t	Coefficient	t
Lag 1	-2.8451	-2.50	0.5862	0.48
Lag 2	-2.5848	-2.28	2.7270	2.81
Lag 3	-1.8172	-1.60	3.6832	3.95
Lag 4	-0.9385	-0.82	3.8831	4.18
Lag 5	-1.2352	-1.09	3.6880	3.91
Lag 6	-1.8579	-1.66	3.6762	3.87
Lag 7	-0.8954	-0.81	3.2089	3.45
Lag 8	-1.0244	-0.91	2.8579	3.15
Lag 9	-1.0573	-0.95	3.2250	3.54
Lag 10	-0.7940	-0.71	2.9692	3.21
Lag 11	-0.2474	-0.22	2.3015	2.44
Lag 12	0.1876	0.17	1.4644	1.58
R 2	0.2445		0.5553

Monthly TR time-series are used from March 1998 till August 2005. Equity spot market volumes are summed across NSE and BSE. In the case of the bond market, turnover recorded at the RBI is used. Returns on the equity market are proxied by monthly Nifty returns. Returns on the bond market are proxied by returns on the notional 10-year bond off the NSE zero coupon yield curve. These tables present OLS models to explain the TR in a month against lagged returns over the last 12 months. Both returns and TR time-series are I(0).

When interest rates turned, the bond market liquidity dropped back to levels prevalent as of six years ago. This suggests a lack of resilience in bond market liquidity.

We can obtain some insights into resilience by measuring the relationship of the TR upon lagged returns. As Table 3 shows, previous values of Nifty returns have a significant effect on stock market TR for two months. However, these seem to be contrary to our expectation: the equity market TR is actually higher following months of negative Nifty returns. On the bond market, current liquidity is dependent on bond market returns from much longer in the past. Here, the coefficients are all positive, which means that higher

ISS Three-month diploma programme UNIVERSALISING SOCIO-ECONOMIC SECURITY FOR THE POOR January 10-March 23, 2007

Applications are invited for the diploma offered by the Institute of Social Studies in The Hague, The Netherlands, in co-operation with the Self Employed Women’s Association, Ahmedabad and the Centre for Development Studies, Trivandrum. Limited Fellowships are available for applicants having professional experience relevant to the programme. For admission procedure and fellowships: Visit: www.iss.nl, email: student.office@iss.nl, Or www.cds.edu Write: Student Office, Institute of Social Studies, P.O.Box 29776,2503 LT The Hague, The Netherlands. Tel: +31 70 4260 Fax: +31 70 4260 799

Closing date for application to the ISS: 11 September 2006 For fellowships: 1 November 2006 for others. For more information: email: messkoub@iss.nl

turnover is associated with higher bond market returns, and vice versa. These results suggest that bond market resiliency is weaker than that of the equity market.

Granger causality testing in these two problems, with 12 lags, shows that there is one-way causality from 10-year bond returns to bond market TR, at a 99.9 per cent level of significance. In contrast, on the equity market, there is no causality in either direction between Nifty returns and the equity market TR.

Conclusion

Liquid securities markets are a core goal of financial sector reforms, and this has two dimensions: low impact cost and high resilience. Three elements of evidence suggest that equity market liquidity is robust to negative price shocks: the experience with a large drop in Nifty on May 17, 2004, the experience with a large drop in the price of Infosys on April 10, 2003, and the regression explaining equity market TR based on lagged Nifty returns. Conversely, bond market liquidity appears fragile when faced with negative price shocks, as seen in the regression explaining bond market TR based on lagged returns on the notional 10-year zero coupon bond.

EPW

Email: susant@igidr.ac.in

[The views expressed in this paper belong to the author and not her employer. I am grateful to CMIE and NSE for the data used in this paper.]

Economic and Political Weekly August 12, 2006