C-3 Means, Standard Deviations, and Other Statistics (Sample B) 97 C-4 Correlations (Sample B) 98 Eigenvalues (Sample B) 102 C-6 Rotated Factor Matrix (Sample B) 103 C-7 Residual Correlations after Removal of 18 Factors (Sample B) 105 C-8 Means, Standard Deviations, and Other Statistics (Sample C) 109 C-9 Correlations (Sample C) 110 Residual Correlations after Removal of 18 Factors (Sample C) 117 Average Percent of Daily Dollar Volume Not Yet Cleared 128 D-5 Estimates of Calendar Days Required for Clearing Various SUMMARY In the summer of 1968, Arthur D. Little, Inc., was invited by the National Association of Securities Dealers, Inc. to assist it in understanding and reducing the fails problem in over-the-counter stock clearing. Early in that work ADL suggested that the performance of the stock clearing system in the United States be reviewed to discover as nearly as possible the causes of fails, thereby guiding the conception and development of remedies. This report is a description of that diagnostic review, of its findings, and of the resulting guidelines by which proposed remedies for the stock clearing problem should be judged. The diagnostic review began by asking knowledgeable people in the industry - in the exchanges/associations, broker/dealer firms, stock clearing houses, and banks - what they thought were the causes of fails. We were soon convinced that there were potentially multiple causes of fails and, in September and October, conducted two surveys of trades which had occurred in August and September 1968. One survey examined the inter-city and inter-regional character of over-the-counter trading and clearing. Another survey, much the larger portion of the diagnostic work, examined about 2100 buys which had occurred in 20 different broker/dealer houses in 6 principal cities. A total of 57 different conditions affecting each buy were recorded and examined for their influence on each other and on the time at which the buy was settled. The survey was done sampling buys made in the New York exchange, American exchange, and regional markets as well as in the over-the-counter market, and sampling buys originating in different parts of the United States, because the conditions affecting stock clearing and the availability of stock clearing houses' services varies significantly for the different markets and geographical locations. The survey was designed to learn how these different conditions affect stock clearing performance. SIZE OF THE STOCK CLEARING PROBLEM Findings confirm general industry knowledge that the stock clearing systems are performing well below the industry standard which indicates that settlement should be made on the fifth business day following the moment of trade. Findings also show, often for the first time, some quantitative indicators of the performance of the stock clearing systems in the United States. In August and September 1968, about 65% of the buys made on the New York Stock Exchange were being cleared on settlement day, about 45% of the buys on the American exchange were clearing on settlement day, and about 20% of the buys in the over-the-counter market cleared on settlement day. It took about 27 calendar days for 99% of buys made on one day on the New York exchange to clear, about 60 days to clear 99% of American trades made in one day, and about 78 calendar days to clear 99% of the over-the-counter buys made in one day. It was estimated from the ADL survey data that the total dollar value of fails-to-receive for the New York, American, and over-the-counter markets combined on a typical day in September 1968 was $4.4 billion, a figure that is somewhat higher than the April-June levels estimated by the New York Stock Exchange from its June 1968 survey data. Examination of the definition of a "fail" indicates that foregoing estimates of the dollar value of fails-to-receive are lower than those which would be made if uncompared trades - trades on which the selling broker's and the buying broker's records of the trade do not compare perfectly - were included in the definition of a fail and in the survey data. The meaning of the current performance of the stock clearing systems can be understood in an historical perspective if it is recalled that "settlement day" in the United States was moved in 1946 from the second business day following the moment of trade to the third day, from the third to the fourth day in 1952, and from the fourth to the fifth day in 1968, probably indicating a general slowing of stock clearing performance over two decades. The ADL surveys also show that approximately one-third of all over-thecounter trades are made between broker/dealers located in different regions of the country. Forty percent or more over-the-counter trades are made between broker/dealers located in different cities, indicating that 40% or more of the trades in over-the-counter securities are cleared by the stock clearing mechanisms which are used for inter-city and inter-regional clearing. It is likely that roughly similar inter-regional and inter-city trading patterns exist for listed securities, although this was not observed in our surveys. The meaning of this finding is clearer when it is recalled that all inter-city and inter-regional stock clearing must be done on a trade-by-trade basis through the banks or by mail. These methods of inter-city stock clearing allow no reduction of stock clearing work that might be realized by inter-regional netting, thereby potentially reducing the inter-regional flow of stock certificates. THE MULTIPLE CAUSES OF FAILS Examination of many conditions affecting or potentially affecting the settlement of a buy as well as the actual time required for the buy to be settled leads us to the following rank-ordering of the causes of fails in the stock clearing system as it was operating for New York, American, and over-the-counter trades in August and September 1968. The most important influence on the time required for a buy to clear is the stock clearing performance of the two parties to the settlement. Both the buying broker/dealer and the broker/dealer making delivery of the certificates influence the probability of a fail. This influence is larger than, and independent of, the other influences and non-influences which are listed below. The next most important influence on the probability of a fail is the presence of a stock clearing house in the system used to clear the trade. Stock clearing through a clearing house is done more promptly than is direct clearing. This influence, too, is independent of all other influences. Most of the industry-guessed causes of fails had little or no influence upon the probability of a fail or the time required to clear in the stock clearing system as it was operating in August-September 1968. In effect, the total stock clearing system was operating so much slower than the industry standard for the reasons described above that most guessed causes of slow clearing simply did not contribute to a further slowing of the system. Transfer agent promptness in transferring title appears to have Recent fails netting, as done by the stock clearing houses in The geographic distances over which settlement must occur Characteristics of the security such as having been issued The size of the settlement, measured in either dollars or shares, The level of recent trading volume on the exchanges immediate- The general reputation of a broker/dealer is unrelated to his Some of the conditions which did not influence the time required to clear a trade in September 1968 are likely to emerge as influences in a stock clearing system which operates with a shorter time cycle than was being achieved late in 1968. If stock clearing is speeded until it is 95% complete on settlement day, these latent causes of fails may appear. In our judgment they are... the geographic distance over which settlement must be influences which reduce the pool of certificates from The definition of a "fail" so that uncompared trades are included - as must happen in a responsible, industry-wide view of the performance of the stock clearing system – will introduce yet another latent cause of slow stock clearing performance which is... ... accuracy of data capture at the moment of trade. Uncom- The findings about influences and non-influences on the performance of the stock clearing system in 1968 were not expected. Most of the opinions about the causes of fails which were offered by knowledgeable people in the industry were not supported, and one commonly held opinion was even reversed. At the time we started work there was relatively little quantitative information available about the performance of the stock clearing system. We conclude that this lack of knowledge must be corrected if there is to be any important change in stock clearing performance. |