If partial deliveries are made whenever possible, additional costs are incurred for the increased number of deliveries and the extra bookkeeping involved. Based on service bureau charges, the industry cost for the bookkeeping required due to the additional deliveries was estimated as $6 million per year.

When it was specified that partial deliveries would be made to brokers and institutions, the cost of fails was reduced from $125 million to $79 million, most of the savings resulting from fewer broker-to-broker fails. Of the various individual changes tested, making partial deliveries appears to be one of the most promising areas for net cost reduction.

DKs, Uncompares, and Wrong Denominations

The combined effect of reducing DKs, uncompares, and wrong denominations was also examined. They were examined in combination because they have the common characteristic of causing delays in trade completion for reasons other than the lack of stock certificates.

DKs affect fails-to-deliver to institutional customers. They do not directly affect deliveries to brokers or cash customers. In the model, when a DK occurs, the stock used in the attempted delivery is held by the delivering broker until either the receiving bank obtains the necessary instructions or the irregularities preventing the delivery are corrected.

Since the bulk of DKs are caused by lack of instructions, transmission of instructions by an interbank wire system instead of by mail would greatly reduce this type of DK. The standard telegram charge was used to estimate an annual cost of $6 million incurred to reduce this type of fail.

A wrong denomination means that stock must be sent to transfer in order to obtain certificates in smaller numbers of shares, e.g., a 1000 share certificate is broken down into a 100, two 200, and a 500 share certificate. During the time it is in transfer, it is removed from the inventory available for making deliveries.

In order to reduce the incidence of wrong denomination problems, brokers need an inventory of stock certificates of various denominations with continuous monitoring so that advance action can be taken to obtain certificate sizes that have been depleted. Therefore, only those firms with a large inventory can readily alleviate the wrong denomination problem by proper inventory management. It was estimated that the addition of at least one employee for each of the fifty largest firms at a total annual cost of less than $1 million would be necessary to have an impact on this problem.

An uncompared trade has a somewhat different impact. Until the two parties to the transaction agree to its terms, no broker-to-broker delivery is made and none is attempted. There will be a failure to complete the transaction until resolution can be accomplished.

The level of uncompared trades for stocks listed on a major exchange can be reduced by a system like the AMEX's proposed Floor Derived Clearance System. For a similar system, it was estimated that the cost would be about $5 million per year for the NYSE.

The combined effect of lowering the DK rate from 30 to 10 percent, the uncompare rate from 6 to 1 percent, and the wrong denomination rate from 5 to 2 percent is a reduction in net cost of $59 million per year-from $125 million to $66 million.

Although each of these factors has an impact on system performance, the greatest benefit would be derived by reducing the DK rate and, hence, the level and cost of broker-to-institution fails. The annual cost of achieving improvements in these three areas would be small ($12 million per year) in relation to the resulting saving.

Combination Changes

The cases discussed up to this point include one change at a time from the benchmark case specifications (DKs, uncompares, and wrong denominations being considered a single type of change). Making two or more changes at the same time produces results that reflect their interaction and, of course, will yield greater savings. Rand has examined a number of different combinations of which two examples are presented here one involving changes only in broker operations and the other involving both brokers and banks. These are identified as minimum structural changes in order to distinguish them from the more far-reaching changes involved in certificate depositories and nationwide clearing for OTC stocks.

Minimum Structural Change—Brokers Only. The purpose of this combination was to explore the effects of system changes that did not assume bank action to reduce transfer time and the DK rate. Brokers were assumed to (1) take full advantage of the partial delivery rules, (2) give first priority to broker deliveries, and (3) use stock loans to cover fails to institutions. This combination reduced the cost of fails from $125 million to $89 million for a net saving of $36 million. This is a substantial saving but still much less than is possible if brokers' improvements are combined with banks' reductions in transfer time and the DK rate. This is shown in the following case.

Minimum Structural Change-Banks and Brokers. For this combination case it was assumed that both the brokers and the banks modified their operating procedures by-(1) fully implementing the partial delivery rules, (2) giving first priority to deliveries to institutions, (3) reducing transfer time, and (4) reducing the rates for DKs, uncompares, and wrong denominations. After allowing for the cost of these changes, the net cost of fails would be reduced from the benchmark total of $125 million to $36 million, a net saving of $89 million for trading in NYSE-listed stocks alone.

Stock Certificate Depositories

To explore the impact of certificate depositories on the trade completion system, two cases were examined, one identified as a limited and the other as a full

depository. The basic difference between the limited and full depositories is the extent to which banks and customers participate. For details on the differences between the two depository cases, see items 11 and 12 in Table 1(page12). It should be noted that the improvements included in the preceding case (minimum structural change-banks and brokers) and shown as item 10 in the table are also applied in the two depository cases, along with additional changes intended to reflect depository operations.

Limited Depository. The limited depository case represents current plans for CCS development. It was assumed that brokers participated fully, that major New York City banks participated to the extent of receiving and delivering stocks, and that institutional and cash customers' holdings outside the depository remained the same. Specifications for this case include mandatory stock loans to cover broker-tobroker fails, first priority for delivery to institutions, full implementation of the partial delivery rules, fast transfer and reduced rates for DKs, uncompares, and wrong denominations.

The cost of all the improvements included in the limited depository case (item 11 in Table 1) was $50 million of which $20 million was for the depository operations alone. The balance of $30 million covered the costs of improvements in transfer time, delivery priorities, and DK, uncompare, and wrong denomination rates. It was assumed that the depository would facilitate partial deliveries without additional cost to the brokers. The net effect was to reduce the cost of fails from $125 million for the benchmark case to $30 million for the limited depository-a saving of $95 million.

The $95 million saving for the limited depository is not much greater than the saving of $89 million found for the banks' and brokers' minimum structural change case (items 10 and 11 in Table 1). The added saving due to the limited depository operation was small primarily because the industry incurs additional costs for the new system at a time when it is not able to obtain the saving of a full-scale depository that would relieve the brokers of most of the costs of handling stock certificates.

Full Depository. To test the implications of a full depository it was assumed that 85 percent of the institutional and cash customers would leave their stock certificates in the depository. Also, it was expected that the benefits to be derived from faster transfers, partial deliveries, stock loans to cover broker-to-broker fails, and lower rates for DKs, uncompares, and wrong denominations would be achieved at lower cost or to a greater degree through an effective and comprehensive depository. In addition, there are substantial added advantages when bookkeeping entries replace actual certificate deliveries.

The incremental costs attributable to the operation of the depository are $51 million, which includes costs for depository operations ($20 million), maintaining additional customer accounts ($20 million), reducing uncompares ($5 million), and reducing DKs ($6 million). Some of the improvement costs are not itemized since they are included in total operating costs for the depository.

The net cost of fails for the full depository is $53 million before allowing for the reduction in brokers' operations department costs that would result from the substitution of bookkeeping entries for the physical movement of stock certificates. It is estimated that savings in cage operations alone could amount to $36 million per year. Subtracting that amount reduces the net cost of fails for the full depository alternative to $17 million per year (item 12 in Table 1) in contrast to the benchmark case's $125 million.

Comparison of OTC Clearing Systems

All available data indicate that the dollar value of fails in relation to trading volume is greater for OTC securities than for listed securities. Experience with the NOTC clearing operation in the New York area shows that, at least locally, substantial improvements can be obtained by organizing a clearing function for OTC securities. Approximately 35 percent of OTC trades were being cleared through a clearing facility and 65 percent through the mails at the time Rand compared the costs of different clearing systems for OTC stocks. These percentages were obtained from NASD.

The cost of implementing OTC clearing systems includes the cost of clearingrelated back office operations, clearing, and the cost of settlement because, when securities are shipped draft attached, settlement costs are quite significant. Back office costs and settlement charges were estimated on the basis of data obtained from a number of OTC firms. Information was obtained from the NOTC (daily net) on clearing and settlement charges, and from the Pacific Coast Stock Exchange on charges for CNS. The cost of fails associated with this existing mode of clearing is estimated to be $91 million per year (at an assumed average OTC daily trading volume of 13 million shares).

If the daily net clearing concept of NOTC were expanded to include the bulk of the large trading firms outside New York City, then the net cost of fails would decline to $65 million per year for a saving of $26 million annually. However, the maximum saving will occur when the CNS clearing and settlement system of the National Clearing Corporation goes into effect. It is estimated that the saving, net of the cost of improvements, will be $52 million per year over the current methods when this system is fully operational.

SUMMARY

The cases covered in the preceding discussion are summarized in Table 1, page 12. Each case is identified by a number and a brief title in the columns at the left of the table. The upper part of the table (items 1 through 12) pertains to trade completion alternatives for NYSE stocks; the lower part of the table (items 13 through 15) pertains to alternatives for OTC stocks with emphasis on different clearing systems Under the general caption "Inputs" are shown the assumptions

made for the benchmark case and the changes made for each of the alternatives tested. A blank space in any input column means that the input for the benchmark case was also used for the alternative case. Whenever an input for a particular case was altered, the substitute input appears in the table. For example, a DK rate of 30 percent for broker-to-bank deliveries was specified for the benchmark case but was reduced to 10 percent for the alternative in which fail rates were changed (item 8). Under the caption "Total Net Cost" the total clerical and interest costs incurred by brokers because of fails plus the net costs of modifying the trade completion system are shown. The last column shows the difference in total net cost due to the changes identified. For the NYSE stocks, the total net cost for each case is compared with the benchmark cost of $125 million. For the OTC cases (items 14 and 15), total net costs are compared with the cost of $91 million for the existing OTC clearing system (item 13).

As the summary table and the preceding discussion show, some alternatives are much more advantageous than others. Action to improve the trade completion system should begin with those changes promising the largest benefit. This should not be interpreted to mean that the less promising changes should be ignored or deferred simply because the saving potential is less. Even though the benefits of combined changes will usually be less than the sum of the savings resulting from single individual changes, the combination changes will yield the largest benefits. For example, the alternative identified as the minimum structural change-banks and brokers (item 10)-yields a saving almost as great as that for a limited depository (item 11). A fully developed depository will provide an additional saving. The point to be emphasized is that significant improvements can be made either independently of a depository or embodied in a depository; the improvements analyzed are consistent with the depository concept and would not have to be cancelled when a depository is fully implemented. Also, development of a nationwide clearing system for OTC stocks does not mean that the benefits shown for the test cases are no longer applicable. Some improvements can be made within a relatively short time and then incorporated with a full depository or OTC clearing system when they become fully operational at a later date.

It is important to recognize that the capability to evaluate a wide range of alternative trade completion systems has now been created. We have demonstrated the usefulness of this capability by applying it to a number of possibilities that seem important. There are without doubt many more that should be and now can be considered in similar fashion. A number of these are outlined in the next section.