DATA ABOUT THE SECURITIES AND TRADING ENVIRONMENT

The 2100 buys included a total of 1030 different securities. Data about the individual securities came from published sources, primarily the National Quotation Bureau, Standard and Poor, and the daily newspapers. Data about total trading volume on the exchanges came from daily newspapers.

Data about recent issues, stock splits, tender offers, and dividend dates were supplied by the National Association of Securities Dealers, Inc. with the help of Financial Information Services, Inc., New York, for some of the data.

Several hundred market makers in over-the-counter securities were asked to estimate for each of several OTC securities the average daily volume, average daily short sales, and floating supply in August 1968. Many responded with estimates, but the data were not complete enough to allow this information to be used in the analysis.

Broker/dealers were given a list of securities and asked to rate "the promptness with which title to stock has been transferred (in your firm's experience) during the last six months." The firms supplying these ratings in time for use in our analysis were:

Ratings also were received from a firm in Houston and a firm in San Francisco, but the data arrived after the complex computer assembly of data was well advanced and therefore could not be used.

DATA ABOUT BROKER/DEALERS

The 2100 buys involved a total of 487 different broker/dealers as buying broker, selling broker, and/or settling broker. Data about these broker/dealers were supplied in large part by the National Association of Securities Dealers, Inc. from their records and from industry sources. Several special data gathering efforts regarding broker/dealers were undertaken for this study.

Stock clearing houses were asked to rate broker/dealers on "the accuracy with which (broker/dealer members) have kept their records of trades and settlements during the last six months." The stock clearing houses supplying these ratings were:

27. New York Stock Clearing Corp. 28. Stock Clearing Corp. of Boston

29. Pacific Coast Stock Exchange Clearing Corp.

New York

Boston

San Francisco

Broker/dealers were asked to rate other broker/dealers on "the promptness with which settlement usually occurs when a broker/dealer has had a stock delivery to make to your firm during the last six months." The broker/dealers supplying these ratings were:

30. Kidder, Peabody & Co., Inc. 31. Shearson Hammill & Co., Inc. 32. Burgess & Leith

33. Rotan Mosle-Dallas Union, Inc. 34. J. Barth & Co.

New York

New York
Boston
Houston

San Francisco

The National Association of Securities Dealers, Inc. provided a rating of broker/dealers' fails-to-deliver positions based upon a January 1968 survey of the membership.

Examiners of the National Association of Securities Dealers, Inc. rated broker/dealer performance "in paying their financial obligations promptly during the last year." These ratings primarily identified broker/dealers who had experienced financial problems which came to the attention of the National Association of Securities Dealers, Inc.

APPENDIX B

SURVEY SAMPLES

Arthur D. Little, Inc., conducted two surveys in September-October 1968 with the help of many cooperating firms (see Appendix A).

Survey 1

Survey 1 was a survey of about 2100 buys observed in 20 houses located in six principal cities. The buys were made primarily in August and September 1968 and were observed in the records of the buying broker's firm in October 1968 by ADL staff. The survey's purpose was to diagnose the multiple causes of fails in stock clearing and their relative contributions to the fails problem for securities of all kinds with special emphasis on over-the-counter securities. Survey 2

Survey 2 was a survey of about 5500 trades
observed in 18 houses located in eight principal
cities. The trades were made in September 1968
and were reported to ADL by the participating
firms. The survey's purpose was to describe the
patterns of inter-city and inter-regional trading
in over-the-counter securities.

In addition to these surveys, the work has included interviews with people in many segments of the industry and country. It has also included the examination of documents and reports. This appendix describes in particular the use of the data from Surveys 1 and 2.

Survey 2 data are used in this report in Table 4 where they describe statistically the inter-regional trading as experienced by the 18 houses participating in the survey. This is the only use of Survey 2 data in this report.

Survey 1 resulted in the observation of 2114 buys which were divided into a variety of sub-samples for various purposes and other constraining reasons. Descriptions of the sub-samples and of their purposes follow.

Sample A N = 210

This sample was chosen to investigate the effect
upon fails which the seller may or may not have
when the settlement is being done through a
stock clearing house. Usually the seller or selling
broker is not the settling broker because the
stock clearing house writes the balance order in-
dicating the delivering broker or settling broker
without regard to the identities of the sever-
al selling brokers involved in the original trades.
The settling broker is selected on the basis of the
number of shares in his net delivery position for
the security being balanced for the day and is
matched with a buying broker so that deliveries
and receives match net positions for both parties
for that day. The sample of 210 buys studied in
Sample A consisted of 70 buys each of over-
the-counter trades, NYSE trades, and ASE
trades all of which had been cleared through a
clearing house and each of which had complete
data on the 39 variables being studied, primarily
the characteristics of selling and settling broker/
dealers. While the analysis showed that the char-
acteristics of both the selling and settling bro-
ker/dealers were correlated with the number of
days required for the balance order to be settled,
we judged the correlations between selling bro-
ker characteristics and days in fail state to be
explained by influences other than the direct
action of a selling broker on the settlement of a
balance order to which he was not a party.
Logically the selling broker cannot directly
influence the settlement of a balance order to
which he is not a party. Therefore the observed
correlations must be the consequence of other
more indirect influences.

Samples B and C were prepared for the major analysis of the data from Survey 1. They were developed using identical operations for their selection.

Mic B N = 439 Sample CN= 440

Buys with complete data on 86 different variables (out of 121 originally planned for study) were selected from the 2114 buys for inclusion in Samples B and C. Altogether 879 buys satisfied this condition. An additional eight variables were retained for study, whether or not observations had been made for these variables, making a total of 94 variables available for study in the analysis based upon Samples B and C. The 879 buys were ordered by transaction number which corresponded to the order in which the buys were observed, and this sample was divided into two samples with the first buy being assigned to Sample C, the second to Sample B, the third to Sample C, and so on until all of the 879 buys were assigned.

The analysis of data from Samples B and C was done using 57 variables, and is described in Appendix C. The 121 variables selected for study were reduced to 57 variables for the major analysis in the following ways:

35 variables were dropped from consideration because of the large incidence of missing information for these variables.

18 variables describing the selling broker/dealer were dropped from the analysis on the basis of the logical and statistical conclusions following examination of data from Sample A.

11 variables were dropped from the analysis of Sample B, and subsequently were also dropped from the analysis of Sample C, because (a) they correlated very highly with other variables which were retained in the analysis, usually correlating in the range from 0.75 to 0.98