APPENDIXES APPENDIX VI-A COMPUTER ANALYSIS OF NYSE MEMBERS' TRANSACTIONS-METHOD OF STUDY, TABLES AND CHARTS To study the trading of the various classes of NYSE members the Special Study undertook to collect data for all members' trading for a period which was long enough to be representative but brief enough to present a manageable reporting burden for member firms. Accordingly, the 15 business days included in the weeks ending January 27, March 24, and June 16, 1961, were chosen. During these 3 weeks the Standard and Poor's "500" Stock Composite Index changed +1.28, -0.18, and -1.48, respectively. For this period data on member trading were collected by the means described in parts D, F, and G of this chapter. While data were collected for all stocks, because of differences in characteristics of trading in preferred and guaranteed issues and the form in which much of the data was available the analysis was made of the common stocks only. Though limited to trading in common stocks in 15 business days, the data were so voluminous as to be susceptible to detailed analysis only through the use of advanced data-handling techniques. This aspect of the analysis was carried out with the aid of the Computation Laboratory of the National Bureau of Standards, using a 7090 IBM computer. As a basis for analysis of this material, market statistics for each stock day were obtained in IBM-card form from the firm of Merrill Lynch, Pierce, Fenner & Smith, Inc. This information, prepared originally by the NYSE Stock Clearing Corp., contained the high, low, and closing prices, the previous day's closing price, the number of transactions, and the total market volume of shares cleared for every stock traded on the NYSE during each business day of the 3 weeks studied. The number of shares cleared through the Stock Clearing Corp. was used, partly as a matter of convenience and partly because its coverage is more comprehensive than the only existing alternative, volume reported on the tape. As discussed in parts D and J, members are not required to report all transactions for inclusion on the tape, and in fact for the years between 1937 and 1961 tape volume has been as much as 9.6 percent below total round-lot volume (in 1946), although in recent years the difference has been around 4.5 percent (table VI-48). A comparison made of daily tape and cleared volume in every 10th common stock during the 3-week period indicated that in 43 percent of the total cases tape volume was identical to cleared, and in another 43 percent differences were smaller than or equal to either 5 percent or 300 shares. Where differences did occur, cleared volume was greater than tape volume in two out of three cases. Although neither available source of volume by stock was wholly satisfactory, use of clearinghouse data avoids overstating the degree of members' participation in trading. The unit of time and of trading utilized for analysis of this material was the single stock on a single day, referred to hereinafter as a "stock day." This was the most refined unit for which complete market data were available. On this basis, member trading was examined in each individual stock each day and related to price developments in the individual stock on that day, rather Trading data for odd-lot dealers were obtained for the same period directly from the odd-lot firms. Such data paralleled those obtained for members' off-floor trading as set forth in pt. G. 2 "Total round-lot volume" represents aggregate data reported by member firms and published in the Commission's weekly statistical releases and monthly bulletins. Since this information is not segregated by stock, it could not be used for detailed analysis in this study. On an annual basis, aggregate cleared volume exceeds tape volume because of omissions from the tape, and for other reasons. than to market price averages. For example, members' net balances for a particular day-purchases minus sales were calculated for the individual stock day and related to the percent price change for that stock day-the difference between the present and previous day's closing prices as a percent of the previous day's closing price in that stock. All of the applicable measures described below were calculated for each class of member for each stock day.4 To study the pattern of relationships for each of the several thousand stock days in which a class of members traded during the 3-week period, each of these stock days was placed in a two-way frequency distribution. For example, to study the relationship between price change and all members' daily balance, the following steps were taken: all members' balances for all stock days were ranked, from the greatest sale to the greatest purchase balance, and 10 ranges were established with about the same number of stock days in each. The same procedure was followed to arrive at 10 price-change ranges with about the same number of stock days in each. With one set of ranges listed vertically and the other horizontally, a 10-line-by-10-column grid was constructed, with each of the hundred resulting cells representing a combination of one range of price changes and one range of members' daily balances. Each of the several thousand stock days with member trading was then assigned to the cell corresponding to the members' net balance and the price change for the stock day. The resulting table showed the number of stock days in which members registered daily net balances of an indicated size range while the price of the stock changed by an indicated range for the day. To adjust for the fact that all horizontal and vertical ranges were not exactly equal in the number of stock days they contained, the table was converted to a percentage basis. The number of stock days in each cell was calculated as a percentage of all stock days in the particular range of, for example, price change; that is, the number of stock days in each cell was converted to a percentage of the total number of stock days in that horizontal line of cells. (See table 9, the appendix.) This percentage distribution table was then converted to graphic form so that any patterns or relationships could be more clearly seen. For this purpose a chart form was developed which uses shadings from white to black to indicate the relative distribution of stock days among the particular horizontal lines of cells. Where a high percentage appears in a cell, the corresponding cell in the chart has been shaded black; where there is a very low percentage, the cell is left white. Intermediate shadings between white and black have been used to indicate percentage groups between the extremes. The key to which percentages correspond to each shading is supplied at the right of each chart. The chart may also be regarded as a visual representation, with darker shadings showing the presence of higher percentages (of line totals) in those cells. so that repeated reference to the key is not necessary. A tendency for darker shadings to appear in cells of larger sales balances on those lines with greater positive price changes, for example, would appear in the chart as a clustering of darker shadings along a belt moving downward and to the right; this pattern would indicate that the class analyzed in the chart tends to be a larger net seller in stock days with greater price increases. (See chart 9, this appendix.) This type of shaded chart, showing the frequency distribution of stock days in percent of line totals for various measures of trading and price, is the primary means used for presenting the results of the stock day analysis. The various trading and price measures used in the analysis fall into several categories. Market characteristics reported or calculated for each stock day were as follows: 1. Market volume of sales. 2. Number of market transactions. 3. Closing price for that day. 4. Price range, measuring the difference between the high and low price for the day, taken as a percent of the closing price. 5. Price change, measuring the difference between the closing price for the day and that of the previous day, taken as a percent of the previous day's closing price. 4 The concept of the stock day is explained also in a footnote which appears in pts. C, D, and F of this chapter. In addition, several measures of trading by all members and by each member class were used: 1. The sales participation rate, comprising the member class' sales volume as a percent of market volume of sales in the stock day. 2. The purchase participation rate, comprising the member class' purchase volume as a percent of market volume of sales in the stock day. 3. The participation rate, comprising the member class' purchases plus sales as a percentage of market volume of sales in the stock day. Since for all classes except specialists two members of the same class may be on both sides of a transaction as principal, it is possible for the class' purchases plus sales in a stock to be greater than the total market volume of sales. While the sales participation rate and purchase participation rate may each only go as high as 100 percent, the total participation rate of each class of members other than specialists can go as high as 200 percent. 4. For specialists and floor traders, the transaction participation rate. comprising the number of the member class' transactions as a percent of the number of market transactions. 5. The balance for the day, comprising the member class' purchases minus the member class' sales in each stock day. 6. For specialists, the closing position for each stock day as a percentage of average daily market volume in the stock over the 3 weeks. 7. Also for specialists, a tick test percent calculated for each stock day." In addition to the analysis carried out on a stock day basis, a separate examination was made of each specialist unit, and of all the common stocks assigned to it, for the 3-week period as a whole. The delineation of specialist units and their stocks raised certain problems of proper allocation. In about 50 common stocks on the Exchange, competing specialists were registered in the same stocks and reported their transactions separately. For the purposes of this study, in these and in some other cases, specialists who reported separately were grouped together as a single specialist unit in the stocks to which they were all assigned. Capitalization of the component specialist units was prorated among the resulting specialist units in the analysis, according to the number of stocks. Other data were also adjusted to take this grouping of some specialist units into account. Several measures for the 3 weeks as a whole were calculated for each specialist unit, as follows: 1. Specialist volume-the total volume of all of the specialist unit's purchases and sales in all of the common stocks assigned to it. 2. Market volume of sales-the total market volume of sales for all common stocks assigned to each specialist unit. 3. Number of specialist's transactions—the total number of specialist transactions in the common stocks assigned to the specialist unit. 4. Number of market transactions—the total number of transactions in the common stocks assigned to each specialist unit. 5. Volume participation rate-the specialist's volume of sales plus purchases stated as a percentage of market volume of sales for the unit's common stocks. 6. Transaction participation rate-the number of specialist's transactions stated as a percentage of the total number of market transactions for the unit's common stocks. 7. Stabilization test versus previous tick--the volume of the unit's sales on plus ticks and purchases on minus ticks as a percentage of the unit's total volume of purchases and sales. 8. Stabilization test versus previous close-the volume of the unit's purchases at or below the previous day's closing price and sales at or above the previous day's closing price as a percentage of the unit's total volume of purchases and sales. 9. Position as percent of market volume-the unit's average daily closing position in its common stocks as a percentage of the average daily market volume of sales of these stocks. 5 See pt. D.6.e (2) of this chapter. Unlike the Exchange's tick test, however, plus tick purchases at prices below the previous day's close were not counted as stabilizing transactions. In the discussion which follows "plus tick" is used to embrace both plus and zeroplus ticks, while "minus tick" is used to mean both minus and zero-minus ticks. 10. Capital, excluding seat value-a calculation of the net capital employed by each unit in the conduct of its specialist business, including cash available for the unit, the value of holdings of assigned stocks at cost, and any credit balance, less the market value of any short position in assigned stocks and any debit balance. This calculation was based on an average for the first and last days of 1960.* 11. Number of common stocks-the number of common stocks assigned to each specialist unit as of December 31, 1961. See pt. D.4.c. |