substantial purchases but larger sales, or few or no purchases and a small volume of sales. Second, by taking the price change over the entire day as its point of reference, this test misses the effect of any significant counter trends which may have occurred during the day. For this reason, though it is generally useful, the test does not necessarily depict the stabilizing quality of specialist trading during the course of the day. Notwithstanding these limitations, the net balance test remains a meaningful measure.

Some of the limitations connected with all these tests reflect the difficulty of measuring price trends either by the moment or the day. Funston has pointed out the inadequacies of using the day as a

measure:

analysis of the specialists' stabilizing activitiy, [based on net balances for the day] furthermore completely ignores the fact that the market seldom moves in an unbroken line, up or down. Suppose, on a particular day, that the market opened down and ran lower for 2 hours. During this time, the specialist bought stock heavily from the public sellers. Suppose, then that during the balance of the day, the trend reversed, and the market climbed slowly but steadily to the close. To the extent necessary, the specialist then became the seller, supplying stock to the public. At the end of day, we might well find that the specialists had bought more stock in the morning's down-swing than they had sold in the latter uptrend. The net result for the day would show a purchase balance by the specialists, with the market higher then the previous day's close.240

The study's analysis has pointed to the inadequacies of the tick test. Greater availability of data about consecutive transactions, aside from other advantages,241 could provide the foundation for the development of more accurate measures of specialist stabilization on the basis of price trends shorter than the day and longer than the period between two transactions. Such a guide could be based on intraday price changes and, while designed and operated with the use of electronic data processing equipment, might very well approximate the commonsense judgment of a price trend that a man on the spot might reach.

With a system registering all transactions, programed tests lodged in the computer could be utilized to alert an exchange official instantaneously to any particular development which may warrant investigation. As plans for the automation of Exchange reporting facilities progress, provision should be made in these systems for floor and market survelliance. The Exchange's stock watch procedure is a step in this direction and, even with limited data, has shown the possibilities of automated data processing in market surveillance.242

(3) Evaluating the stabilization performance of specialists as

a group

The specialists' central role in the market makes it extremely important to determine whether their influence is a stabilizing or destabilizing one. Some light on this problem is shed by the tests for the measurement of individual specialist performance, discussed in the previous section. In evaluating the stabilization performance of specialists as a group, however, problems arise involving the use of aggregate data.

The use of aggregates often submerges significant differences in performance. Aggregate studies not only combine the varying per

240 Funston letter, Harv. Bus. Rev., September-October 1962, p. 7.

241 See pt. J of this chapter.

242 See ch. XII.

formances of different units but also may ignore the fact that on almost any day of extreme price movement, there are many stocks moving against the trend. As noted in the previous section, the Exchange usually states that the specialist tick test percentage was 80 or 85 percent. As also noted, this masks the fact that the tick test percentage on any one day is often considerably higher on the sale side than on the purchase side, or vice versa, and on one side is often considerably different from the combined percentage for both sides (table VI-31).

Possibly the most frequently used overall analysis compares aggregate specialist purchase and sale balances in all stocks with the movement of a price index. The Exchange press release (quoted in section 6.e above) which cited aggregate specialist purchase balances on May 28, 1962, as indicating a high measure of "stabilization" is an example of this test.243 This aggregative test was used by Raymond Vernon in examining specialist trading for the 2 years from March 2, 1936, to March 30, 1938,244 and by Ralph and Estelle James in studying the 2 years from July 1, 1959, to July 1, 1961.245 In both cases it was concluded that specialists more often traded with the price trend than against it.

The Special Study used this technique in an examination of the 497 days between November 2, 1959, and November 1, 1961, on which there were changes in the Standard & Poor's index for the day. Specialists were found to have had net balances with the trend on 280 days and against it on 217 days (table VI-32). The net balances, however, were found to be small when compared with total specialist purchases and sales for the same days during the period, so that no significant conclusion could be reached. The insignificance of these small net balances in connection with stabilization should not lead to neglecting their possible significance in other respects. The aggregate data cited in the next section bear out specialists' testimony that as a group, specialists became bearish near the end of 1961 and therefore tended to reduce their positions for some months prior to the May 1962 break. Continuous study of the aggregate figures would have shown a significant cumulative specialists' net sale balance over these months, and might have alerted authorities to the situation developing and to the impact of specialists' actions.

Another possible application of aggregative analysis may involve separating transactions or stocks by different categories-such as stocks which rose in price and those which declined-and examining the aggregates for each class. This technique, which requires data for specialist transactions in individual stocks, was utilized by the Special Study for the 3-week period in early 1961. Tick test and net balance comparisons were made for 4 days of the 3-week period, chosen to include 2 days of general price increase January 23 and 27, 1961, in which the Standard & Poor's index rose 0.33 and 0.62, respectively-and 2 days of general price decline-June 13 and 15, 1961, in which that index declined 0.35 and 0.29, respectively. On each of

243 But see Funston statement in sec. 6.e(2), above.

214 Vernon, "The Regulation of Stock Exchange Members," pp. 81-89 (1941).

245 James and James, "Disputed Role of the Stock Exchange Specialist," Harv. Bus. Rev., May-June 1962, pp. 136-137. It should be pointed out that the aggregate data relied on in this article and in Vernon's book are the only figures now publicly available. Such data are published by the Commission in its weekly statistical release and its monthly statistical bulletin.

the 4 days, specialists' aggregate balances were with the trend, decreasing on index declines and increasing on index rises (table VI-g).

TABLE VI-g.—Specialists' tick test percentages and purchases, sales, and balances for 4 selected days in 1961

As a further step in the analysis, issues which declined in price were segregated from issues which showed price increases and the purchase and sale balances were computed for each category, for each day. With stocks separated in this fashion, all but one of the eight groups showed specialists' purchase or sale balances against the trend-specialists as a group sold more than they bought when prices rose, and bought more than they sold as prices declined (table VI-h). The different conclusions which thus emerge from disaggregating the same data demonstrate the dangers inherent in combining specialist volumes for all stocks regardless of price performance.

TABLE VI-h.-Specialists' tick test percentages and purchases, sales, and balances for 4 selected days in 1961

[Analyzed by stocks that increased and stocks that decreased]

Using this technique of partial disaggregation for the tick test, it was found that for the 2 days of price increase most of the destabilizing transactions were on the purchase side, and for the 2 days of price decrease most of the destabilizing transactions were on the sale side.

An additional analysis of stocks by direction of price change compared with specialists' purchase and sale balances showed that about one-third of the stocks studied had net specialist balances running with the trend. In these stocks, slightly over twice as many transactions were with the trend as were against it (tables VI-33 to VI-36),246

240 Specialists' purchases and sales in these stocks accounted for 45 percent of total specialist purchases and sales during the 4-day study.

When the tick test was applied to these stocks, they showed the greatest percentage of destabilizing transactions. Moreover, there seems to be a greater concentration of destabilizing transactions that were with the trend. This confirms, to the extent that aggregate analysis can, that balances with the trend were achieved in transactions in which the specialist often "reached" across the market,247 i.e., the specialist did not wait for someone to trade, but rather he initiated the transaction by buying stock at the offer or selling stock to a bid; these bids or offers are often limited orders on the specialist's book.

Though this analysis disaggregated the data only partially, it indicated that while most specialist trading tends to have a stabilizing market effect, a significant group of stocks did exist in which specialists not only accumulated balances in the direction of the price trend but may also have done so in a manner prohibited by the regulatory scheme. These results also clearly indicate the dangers inherent in aggregating and averaging such data. The aggregating of net balances tended to obscure the fact that while most balances were against the trend there were a substantial number of relatively large balances with the trend, and the averaging of purchase and sale tick percentages diluted the usefulness of the test since the percentages on at least one side tend to be high.

Since it is in the individual stock that each specialist's performance is either stabilizing or destabilizing, any evaluation of the impact of all specialsts must rest upon evaluaton of performances in individual stocks.

To determine the effect of specialists' transactions on this basis, the Special Study undertook an entirely different type of analysis. Each stock was analyzed separately for each day of the 3-week study and the results of each analysis were then aggregated. For each of the 16,174 stock days 248 in which trading took place during this period, price movements and market volume were noted and examined against various measures of specialist performance (using a computer).249

It was found that, during the period studied, specialist performance was stabilizing in the large majority of stock days.250 This pattern is evident in several measures used. It may be studied first in appendix VI-A, table 9 and chart 9, which show for stocks in each category of daily price change, the number of cases in which specialists accumulated a net purchase or sale balance and the size of this balance. Generally, specialists tended to accumulate purchase balances in stocks which declined in price for the day and sale balances in stocks which rose, with the balances tending to be greater in size the greater the change in price. However, while the majority showed stabilizing net balances against the trend, a substantial minority showed destabilizing net balances with the trend. Some 27 percent of the stock days with a price rise of over 2 percent for the day showed specialist net purchase balances, one-third of which were over 800 shares.

The stablilizing performance of most specialists was reflected also in the distribution of their total sales and purchases. Chart 10 and

247 It will be recalled that this is the kind of transaction generally condemned by the Saperstein Interpretation. See sec. 6.b, above. 248 The "stock day" concept is explained at p. 85, note 182, above.

249 Only common stocks were included in the analysis. See app. A of this chapter for a further description of this study.

250 This confirms the result in the study just discussed which partially disaggregated trading for 4 days.

table 10, in appendix VI-A, indicates that specialists' sales participation rates tended to increase in stocks which rose in price the most, and to decrease in stocks which declined in price the most. Similarly, their purchases as a percentage of total volume participation rates tended to increase in stocks showing the biggest price decline, and to decrease in stocks showing the greatest price rises (app. VI-A, chart 11 and table 11). Again this stabilizing tendency was accompanied by a significant number of destabilizing performances. In almost 15 percent of the stock days in which prices rose over 2 percent for the day, specialists accounted for more than 25 percent of the total purchase volume for the day, while in about 14 percent of the stock days in which prices declined by over 2.4 percent, specialists accounted for more than 25 percent of the sale volume.

The evaluation of specialist "stablilization" performance as measured by the tick test paints a far rosier picture. Fully 65 percent of the stock days register tick test scores of 100 percent. But even here, in 10 percent of the cases, specialists' scores were 50 percent or less. Moreover, as has been explained in the previous section, the tick test as now administered by the Exchange offers little basis in itself for conclusion on specialists' performance.

Given the evidence presented above, it may be concluded that specialist activity during the period studied tended to stabilize most stocks when each is examined on a daily basis. The presence of a significant amount of destabilizing activity, however, points to the need for developing methods of measuring specialist performance on a more valid and current basis. Only such development, and more affirmative action by the Exchange against specialists whose performance is delinquent, can bring the level of all specialists' performance up to that of the majority.

(4) The market break of May 1962 251

A dramatic test of the degree to which specialist trading may influence market movements occurred at the end of May 1962. Reference has been made to various Exchange releases giving analyses of specialist trading on May 28, 29, and 31, 1962.252 It will be recalled that these releases pointed with pride to specialist activities as "stabilizing" during this period.

The Special Study pursued several lines of inquiry in order to test the possible impact of specialists' dealings during the break. Aggregate specialist trading data for the end of May and for prior months were analyzed. A more refined study was made of specialist trading in 50 issues which were market leaders during the period. Eight of these issues were analyzed transaction by transaction for May 28, 1962. Finally, testimony was taken from individual specialists with respect to their activities during the period of study.

During the period of general market decline from mid-December 1961 until the last few days of May 1962, specialists in the aggregate substantially reduced their positions (chart VI-c). On December 29 all specialists units had a net long position of about 3,600,000 shares. By the close of business on Friday, May 25, they had reduced

281 This section discusses only the activities of specialists during this period; for a study of the 1962 market break, see ch. XIII. 259 See pp. 97 and 98, above.