Stock Market Liquidity: Measurement and Implications

Revised Fourth Capital Market Conference December 20 Stock Market Liquidity: Measurement and Implications M. K. Datar Deputy General Manager Industrial Development Bank of India IDBI Towers, Cuffe Parade Mumbai-45 Tel: 2189111 (Ext: 3030) Fax: 2180411 e-mail : mk.datar@idbi.co.in

2 Abstract Stock Market Liquidity: Measurement and Implications M. K. DATAR The paper focuses on stock market liquidity and its measurement. It discusses relative merits of different measures that have been used for measurement of market liquidity viz. volume/ frequency of trading, turnover ratio and impact cost. It also measures changes in stock market liquidity on the basis of traditional measures such as turnover ratio. The paper proposes Elasticity of Trading as an alternate measure which easy to compute and has superior information content. The elasticity of trading measures volume of trading in relation to changes in prices. The paper analyses liquidity of equity trading on the National Stock Exchange with the help of Coefficient of Elasticity of Trading (CET) during April 18 - May 20. Some directions are suggested to assess the true potential of CET. This could be done by using CET for measuring liquidity in individual scrip. Similarly CET may be computed for period covering trading cycle. The author wishes to thank Dr. R.H. Patil, S/Shri. Anand Shanbhag, S.S. Chandorkar and an anonymous referee for offering comments on an earlier draft.thanks are also due to all participants at the fourth capital market conference where the paper was presented. Needless to mention, the author alone is responsible for errors that may have remained in the paper.

3 Stock Market Liquidity: Measurement and Implications M. K. Datar * Low frequency and small trading volumes along with high fluctuations in prices is a characteristic feature of stock markets in India. Over the last two decades, Indian stock markets have witnessed significant changes in terms of trading environment (particularly the introduction of screen based trading), regulatory framework and higher competition among market intermediaries. In addition, there has been significant increase in the diverse range of instruments offered and informed market players have been playing a larger role. Stock markets have also expanded phenomenally in terms of market capitalisation and funds raised etc. It would be interesting to assess the stock market developments in terms of impact these changes have made on frequency and depth of trading. It may be agreed that unless stock trading is widespread and deep as signified by higher volumes, the beneficial impact of well functioning capital market would remain limited to a few scrips where trading is frequent and deep. This would also result in duel equity markets wherein a small number of scrips attracting large trading interest while large number of scrips show low and infrequent trading volumes. Stock prices are generally treated as lead indicator of future economic activity. This is true provided current prices represent discounted value of expected dividend growth and that, to the extent such assets are traded in deep and well informed markets, expectation about future growth tend to be rational. Stock market liquidity could therefore be an important perspective to assess the stock market developments. This paper discusses alternative measures used for measurement of market liquidity. It also measures changes in stock market liquidity on the basis of certain measures besides suggesting an alternative measure. Concepts of Liquidity One could think of two distinct though related concepts of liquidity. First is monetary liquidity which would depend on availability of cash and near cash in relation to general demand for goods / assets. The trends in monetary liquidity would generally get reflected in short-term interest rates; low short term rates signifying easy liquidity. i This concept of liquidity is commonly talked about and commented upon. In the context of specific markets viz. commodities or equities, depth of market is generally related to volume of transactions or frequency of trading. In popular stock market parlance, volume of trading are routinely indicated to signify ease/frequency of trading. However, to talk about liquidity without reference to price is hardly meaningful. Market liquidity is considered as capacity of financial markets to absorb temporary fluctuations in demand and supply without undue dislocations in prices. ii These two concepts of liquidity are linked as increase in monetary liquidity may lead to higher demand for securities and would reflect in higher security prices. As increase in monetary liquidity would lead to lower interest rates, it is through the increase in security prices the yield would be brought in alignment with interest rates. However, market liquidity should be considered an important indicator of the state of market. A good measure of liquidity across different markets would help in comparative analysis of different markets/ market segments. Measurement of Market Liquidity Attempts have been made to measure market liquidity in terms of volume/ frequency of trading, volume of turnover and impact costs. One simple measure of market liquidity is to measure through frequency of trading. iii More frequent trading would certainly mean improved liquidity but with such an indicator it is not possible to measure extent of liquidity among frequently traded shares. Volume of trading i.e. number of shares traded could also be considered as a measure of liquidity. Gupta (12) has used this measure to detect "excessive" or speculative trading. However, it would be difficult to assess liquidity only with reference to absolute volume of shares traded. A relative * The views expressed are personal and not necessarily of the organisation the author is associated with.

4 measure could be the ratio of traded volume to total number of shares issued which enables comparison across different scrips. However, number of shares actually available for trading are different from number of shares issued because of promoter / strategic or government holding etc. which normally are not traded. As a result, floating stock will be lower than the total issued shares. Adjustment would therefore be necessary to account for this factor while accurately measuring liquidity of different shares. However, such adjustments would be company specific and it would be difficult to do such adjustments ( for arriving at floating stock) at the aggregate level. Moreover, stock prices, anticipated or actual, are linked to demand for stocks and the extent of trading volumes. Hence an ideal measure of liquidity should combine price and volume. One way to combine this is to consider turnover as a measure of liquidity. In fact, liquidity has often been analysed in terms of turnover data. At the aggregate level, trends in annual turnover (i.e. number of shares traded*price) becomes a measure of market liquidity. At times, total turnover in relation to market capitalisation (market price of listed shares) is considered as a relative measure which can be used for comparison across different markets or over time. Table 1 presents the trends in turnover and turnover ratio in BSE trading since 10-91. The turnover ratio displays wide fluctuations. This ratio suffers somewhat from dimensional distortion because while the market capitalisation is a stock measure i.e. at a point of time, the turnover is a flow variable i.e. over a period of time. In the case of individual scrip, the turnover could be compared to market capitalisation of that scrip to get a measure of liquidity that can be used across different scrip. However, the floating stock could be very different in different securities. Several analysts take this factor into account while assessing liquidity in individual scrip. Recently, global stock indices are being recalculated after adjusting market capitalisation of a scrip for proportion of free float. Table 1: Liquidity in Bombay Stock Exchange Trading (Rs. crore) Year Turnover Market Capitalisation Turnover/ Market Cap. (%) BSE Sensex Year end 10-91 36012 90386 39.84 1168 11-92 71777 323363 22.19 4285 (266.8) 12-93 45696 188146 24.28 2280 (-46.7) 13-94 84536 368071 22.96 3779 (65.7) 14-95 67749 435481 15.55 3261 (-13.7) 15-96 563 526476 9.51 3367 (3.1) 16-97 124284 463915 26.79 3361 (-0.2) 17-207644 560325 37.06 3893 (15.8) 18-3119 542942 57.21 3740 (-3.9) 19-20 685028 912842 75.04 51 (33.71) Source: RBI: Handbook of Statistics of Indian Economy and Annual Reports It may be noticed that liquidity as measured by turnover (column 1) and turnover ratio (column 2) display different trends. During 11-92 absolute level of turnover increased but turnover ratio came down. On the contrary, during 12-93, absolute level of turnover fell significantly but turnover ratio improved due to steep fall in market capitalisation resulting from fall in stock prices. Even in 13-94, while turnover improved, turnover ratio fell. During 14-95 and 15-96, both turnover and turnover ratio moved in same downward direction. After 16-97, both turnover and turnover ratio have moved in upward direction. The decline in turnover ratio over 13-96 seems to arise as volumes failed to keep pace with increase in market capitalisation. It needs to be noted that when liquidity is measured in terms of turnover, distinction between volume of shares traded and value of shares traded is lost. Over short periods of time, it may not be material to maintain such distinction but over longer periods it may be useful to do so. Impact Cost

5 Concept of impact cost as measure of liquidity was introduced in Indian markets recently - Shah (16) Thomas Susan (18). It represents improvement over traditional measures like volume / frequency of trading. Basically this measure takes into account institutional features of the screen based trading environment wherein structure of order book would represent liquidity. At any time, order book would have different unmatched sale / purchase orders. From this data, average price to be paid to purchase shares worth (s ay) Rs. 1 mn. can be computed. This is compared with ideal price which should be average of bid and ask price i.e. ( Bid price + Ask price )/2.The % difference between ideal price and required price is termed as impact cost. Lower the impact cost, higher would be the liquidity. There are certain advantages of impact cost as a measure of liquidity. It reflects up to date prices. It could be computed for individual scrip or an aggregate of scrips, say scrips included in Sensex or Nifty. National Stock Exchange uses minimum impact cost as a criterion for inclusion of scrips in stock indices it compiles. It also publishes impact cost for different index scrips averaged over one year period. The problem of combining volume and price is solved by computing impact cost for a standardised transaction. It is of course true that impact costs would vary with transaction size and also over time. Estimates suggest that impact cost during March-July 17 was 0.29% for Rs. 5 mn. worth purchase of index scrips while the same was 0.49% for Rs. 20 mn. transaction. Similarly, impact cost was 0.56% in March 17 while it reached 0.16% in May 17. iv The concept of impact cost is no doubt very helpful for index fund managers as it addresses their concerns when it is computed for sale/purchase of index scrips. Impact cost measures marginal liquidity in a forward looking manner as it computes liquidity of a potential standard size transaction from the order book of an automated order driven stock exchange. It is maintained that liquidity would indicate / cost of trading and lower such costs higher would be the liquidity. But cost of trading would also depend on market structure of and competition in the market. Should liquidity be measured without any reference to volume of trading as done in the case of impact cost? While it takes care of the problem of stale prices, it ignores the dimension of actual volume while measuring liquidity. If liquidity is to denote the ease of trading i.e. volume of trading with minimum impact on price in an instrument / market it needs to be based on volume and price. Moreover, it would be difficult to compute impact costs in traditional market which are not automated and order driven. It would therefore be difficult to use impact cost to compare liquidity in equity markets and markets for corporate bonds or even commodities which have not yet been automated in developing countries like India.. Furthermore, computation of impact cost would be difficult unless one has access to complete order book. It is not easy to compute impact cost unlike say computation of turnover ratio. Therefore, unless the impact costs are computed and published regularly (say by stock exchanges), their use by individual market participants may remain limited and therefore informational role of impact costs would remain incomplete. Elasticity of Trading The proposed new measure to monitor liquidity is similar to price elasticity measure. It is measured as price elasticity of trading volumes. It can be computed for individual stocks or group thereof. It can be computed for any period of time ( day or month ). The main advantage is it can be computed by anyone who has access to information on prices and volumes of trading data. Coefficient of Elasticity of Trading (CET) = % change in trading Volume / % change in Price The range of CET is quite wide : from + infinity to - infinity. The coefficient would be positive when the direction of changes in volume and price is same while it would be -ve when the directions are different. High value of CET would indicate that price changes are accompanied by high volume of transaction. When large sized transactions take place with little or no change in price, value of CET would approach infinity and indicate high liquidity. As a measure of elasticity, the sign of CET may not be very relevant, but with its sign, the information content will be enriched. CET, together with price trends would convey a lot of information about state of the market: either for individual scrip or the whole market. Normally there would be inverse relationship between prices and quantity demanded but in case of

6 asset markets as expected prices have more important role, it would be difficult to envisage any ex ante relationship between prices and volumes. Exhibit 1 : Range of CET Value of CET Price Change Prices increase Prices decline > 1 =1 < 1 Price increases supported by more than proportionate change in volumes. Price declines matched by more than proportionate change in volumes Price increases matched by proportionate change in volumes Price declines matched by proportionate change in volumes Bull Run (?) Bear hug (?) A schematic presentation of the range of potential information is illustrated in Exhibit 1. To illustrate, if CET is above 1 and prices are increasing would indicate a situation where higher prices are drawing even higher volumes and higher prices should reflect real good news. If CET equals 1 and prices are increasing would mean volumes are increasing proportionately. Whereas, if CET is less than 1 and prices are increasing would indicate a situation where prices are increasing on low volumes and the price increase could be speculative. It would be interesting to consider whether CET, as a measure of liquidity, different from other measures of liquidity such as turnover ratio and impact cost? One measure difference is that it is computed on changes in prices and volumes in contrast to turnover ratio which is based on prices and volumes. Computations of impact cost are based on structure of order book and are generally computed for a transaction of standard size. CET and Turnover Ratio The Co-efficient of Elasticity of Trading is computed for NSE trading at monthly rests from April 18 to September 20. The data is presented in Table 4, while the same is visually presented in exhibit 2. For the sake of easy comparison, turnover ratio is also computed for the same period. It may be noticed that these two different measures, at times, display different trends. For example, during whole of 18-, turnover ratio would indicate that, on the whole, liquidity improved generally while the movements in CET would indicate that the liquidity was generally fluctuating and on a steady decline. For the entire period April 18 to September 20 the average value of turnover ratio was 11.5 while standard deviation was 3.14 %. The average value of CET for the same period was -2. and standard deviation was 12.16. Thus these two measures portray quite different pictures about the state of market liquidity. The important question is to decide which reflects the true market condition. Moreover, the range of CET is quite wide and it distinguishes between different price trends which is reflected in the sign of CET. In the case turnover ratio, it is not possible to distinguish between increase in turnover due to increased volumes or increased prices. CET will show such differences in the direction of movements in volume and price in the sign of CET. The richer information content of CET could be illustrated through Exhibit 4 wherein, in addition to the two measures of liquidity, percent change in Nifty for every month is also displayed. As is well known, stock prices displayed a declining tendency till November 18. From December 18, stock prices increased more or less steadily till February 20 except September when prices came down. Prices started declining from March 20 onwards.

7 Exhibit 2: Different Liquidity Measures Turnover Ratio 21. 19. 17. 15. 13. 11. 9. 7. 5. Apr- May- Jun- Jul- Aug- Sep- Oct- Nov- Dec- Jan- Feb- Mar- Apr- May- Jun- Series2 Jul- Aug- Sep- Oct- Nov- Dec- Note: Series 1 is CET: Series 2 is turnover ratio Series1 Jan- Feb- Mar- Apr- May- Jun- Jul- Aug- Sep- 30. 20. 10. 0. -10. -20. -30. -40. -50. -60. -70. Coefficent of Elasticity of Trading The liquidity conditions as reflected by these two measures are quite different during these three periods of different price trends. During April-November 18 as reflected in exhibit 2, the CET was generally maintained in positive range (during 7 months out of total 9 months) except during November and June when prices declined but volume rose leading to a negative value of CET. During the same period turnover ratio increased on 3 occasions while it declined on 4 occasions. Average turnover was 6.69% during this period while average CET was 2.41. Stock prices rose from December 18 and remained firm till February 20. During this period CET was positive and >1 on 6 occasions and negative and >1 on 6 occasions while CET was less than 1 on 3 occasions. This shows fluctuations in liquidity levels. Turnover ratio increased 8 occasions while it declined five times. Average turnover ratio was 9.5 % while average CET was - 4.3%. From March 20, stock prices are displaying declining trend. CET remained at low positive values or large negative values throughout the period. Turnover ratio increased on 4 occasions while the same declined on two occasion. The differences in behaviour of turnover ratio and CET are summarised in Table 2 below : Table 2: Turnover Ratio and CET Period Turnover Ratio CET April 18 - November 6.69 (0.58) 2.41 (7.87) December 18 - February 20 March 20- September 20 Figures in brackets are standard deviation. 9.5(1.34) -4.29 (15.45) 13.17(4.11) -0.71 (6.73) As may be seen from above Table, turnover ratio suggests improvement in liquidity as reflected in average turnover ratio and higher volatility as reflected in std. deviation. CET on the other hand conveys increase in liquidity during the first two periods but reduction in liquidity in the last period. In addition the - sign during second and third period conveys additional information on direction of movements in volumes and prices.

8 CET and Turnover Ratio As indicated above, it is not easy to compute impact cost which makes it difficult to compare impact cost and CET over a long period of time. Shah and Sivaprakasam ( 20 ) have reported impact cost for the period July 97 to July 17. Admittedly this period is rather short for a conclusive comparison. Table 3 presents the data on three measures of liquidity viz. Impact cost, CET and Turn over ratio. An increase in impact cost would indicate decline in liquidity and vice versa. Focus here would be on relationship between Impact cost and CET as comparison between turnover ratio and CET has been done for an extended period above. Exhibit 3 below shows visually the movements in the two indicators. Table 3: Liquidity Indicators: February - July 17 Month Impact Cost CET Turnover Ratio Feb-97 0.21-2.95 6.18% Mar-97 0.56-1.24 7.16% Apr-97 0.21-2.04 4.62% May-97 0.16-5.18 4.86% Jun-97 0.27 2.26 6.78% Jul-97 0.33 4.86 6.94% It may be seen that liquidity trends as measured by CET and Impact cost are similar except for July 17. Impact cost indicates liquidity came down whereas CET indicates an improvement in liquidity. For all other months the liquidity movements are quite similar. While wider comparison may be necessary to consider relative efficacy of these two measures the similarity is quite encouraging. If it is agreed that CET and impact ratio reflect similar trends, CET would have advantage in terms of ease of computation and wider availability of information required. Exhibit 3: Impact Cost and CET 0.6 0.5 0.4 0.3 0.2 0.1 0 0.56 4.86 2.26 0.33 0.27 0.21-1.24 0.21-2.04-2.95 0.16-5.18 Feb-97 Mar-97 Apr-97 May-97 Jun-97 Jul-97 6. 4. 2. 0. -2. -4. -6. IMPCOST CET Prices and Liquidity Trends Is there any relationship between liquidity in and prices of equity shares? As indicated above, it is difficult to hypothesize about relationship between prices and volume of trading. Exhibit 4 presents data on equity prices and liquidity trends on National Stock exchange.

9 30. 20. 10. 0. -10. -20. -30. -40. -50. -60. -70. Apr- M ay- J Julun- A ug- S Octep- N ov- Ehibit 4: Prices and Liquidity Trends D ec- J an- F eb- M ar- Apr- M ay- J Julun- A ug- S Octep- N ov- D ec- J an- F eb- M ar- Apr- M ay- J Julun- A ug- S ep- Series2 Series1 Series3 Note: Series 1 is CET: Series 2 is turnover ratio Series 3 is per cent change in Nifty It is remarkable to note the similarities in movements of series 1 and series 3 i.e. CET and per cent change in Nifty. It is true that in 8 out of 29 occasions the movements in these two series were directionally different. Also, the coefficient of correlation in these two series is statistically insignificant. It is also true that the later is included in the computation of the former. However, visually CET appears to track movements in Nifty. If this indeed is true, it would indicate the peculiar nature of stock market liquidity where volume of trading would be linked to prices. Perhaps this is true of all capital assets, in which case any measure of liquidity should ideally combine prices and volume. Concluding Observations Liquidity conditions provide an important perspective to assess stock market developments. It is important to have a measure of liquidity which can easily be computed on the basis of easily available information. The above analysis indicates that CET may prove be a potentially useful tool to analyse stock market liquidity. While liquidity levels may be fluctuating it may be useful to compute estimates of liquidity with high frequency data ( daily or weekly ) and making use of techniques such as regression. The true potential of CET for analysis of market liquidity may be judged when such measurement is done at individual scrip level. The time interval for which CET is analysed may also be relevant. Perhaps it may be useful to compute CET over a settlement period rather than an arbitrarily chosen period as week or month. Considering the inter-exchange trades, it may become necessary to include volume of trading across different exchanges while preparing security specific measure of liquidity. Such analysis may also be useful for fund managers. The analysis presented above is done for all scrips traded on NSE but the price increases are measured by Nifty index. As it is difficult to have a price index for all scrips, it may be useful to consider volume data only for the scrips included in the index. As regards relative efficacy of CET and Impact cost further detailed work is necessary to arrive at conclusive findings. The comparison for a very short period above indicate broadly similar trends in liquidity. If this is accepted, CET will have an edge in terms of ease of computation vis-vis impact cost. It would be quite interesting to asses the CET across different assets (shares, debentures or gold) and regional markets (national / international). For this purpose CET would be very useful as it is applicable to both modern and traditional markets.

10 Table 4 : Measures of Stock Market Liquidity Month Turnover ratio CET % change in NIFTY Apr- 6.32 3.49 3.80 May- 6.78 1.32-8.30 Jun- 6.96-0.32-11.43 Jul- 6.95 17.09-1.09 Aug- 5.66 1.69-8.44 Sep- 7.36 5.89 6.12 Oct- 7.28 1.66-8.95 Nov- 6.22-11.53-0.76 Dec- 8.20 2.31 8.13 Jan- 9.94 2.61 9.27 Feb- 9.95-16.71-0.51 Mar- 11.67 0.66 12.15 Apr- 8.07 4.50-9.26 May- 9.97 3.82 15.75 Jun- 7.64 0.41 4.89 Jul- 9.34 2.09 10.31 Aug- 8.04-0.26 7.77 Sep- 8.14-56.80 0.08 Oct- 10.78-2.36-8.24 Nov- 9.13-4.54 6.13 Dec- 11.56 3.62 7.61 Jan- 9.03-4.35 4.41 Feb- 11.10 0.61 7.02 Mar- 10.35 2.63-7.64 Apr- 7.09 1.97-7. May- 10.72-15.34-1.86 Jun- 14. 1.79 6.59 Jul- 14.74 0.27-9.42 Aug- 15.78 4.94 4.60 Sep- 19.51-1.21-8.78 Mean 11.53-2. 0.88 Std. Dev 3.14 12.16 7.88 References Gupta L.C. (12) Stock Exchange Trading In India: Agenda for Reforms Society for Capital Market Research and Development, New Delhi. IMF(20) World Economic Outlook May 20 National Stock Exchange NSE News various Issues Shah Ajay (16) The Market for liquidity in Shah Ajay (ed) Future of India's Stock Exchange Invest India Foundation 16 Shah Ajay & Sivakumar Sivaprasam (20) Changing Liquidity in the Indian Equity Market Emerging Markets Quarterly IV(2) Summer Thomas Susan (18) Liquidity Consideration in Index Construction in Waghmare Tushar (ed) The Future of Fund Management in India Tata McGraw Hill Mumbai.

11 i Whether growth in money supply (say M3) or trends in interest rates should be considered as indicator of monetary policy was a widely debated issue among economists. Market participants are, however, are likely to prefer interest rate as easily observable and hence preferred indicator. ii IMF(20) p. 89 iii Gupta (12) iv Shah & Sivakumar (20)