-17

that for $1 million a year the larger firms could each employ at least one full-time person charged with inventory management responsibility and thus reduce the overall wrong denomination problem by approximately one-half. The number of deliveries delayed due to wrong denominations would drop from 5 percent (the estimated current situation) to 2-1/2 percent.

Uncompares.

After a trade has been agreed to by two brokers on the floor of the exchange or over the telephone as is the case in the OTC market, the terms of the trade are reported to their respective operations departments. Prior to settlement, the two departments match the trade, i.e., confirm the fact that the two parties are in agreement about the terms of the transaction. It is estimated that for attempts at comparing through the major clearing houses a mismatch results about 6 percent of the time. It is estimated that the adoption of a system like the AMEX Floor Derived Clearance System (FDCS) would cost approximately $5 million per year including amortization of the cost of installation and would reduce the level of uncompared trades from 6 percent to 1 percent. The major difficulty now is inconsistent reporting from the floor. The FDCS resolves this difficulty by insuring a matching of both sides of the transaction--achieving what is commonly referred to as a "locked-in trade"--before the transaction is reported.

Transfer Time. In the spring of 1969, an average of 8 days was needed to transfer a stock either to broker or customer name. Since certificates are unavailable for trade completion while they are in transfer, reducing transfer time would improve the performance of the system. It is estimated that if each of the 10 major clearing banks (the New York City banks that perform 90 percent of the transfers) were to invest $5 million in automating their transfer function by installing computer driven cathode ray tube consoles for stock record look-up and data entry, they could reduce the average time required for transfer

*

way that the small firms, acting independently, can solve the wrong denomination problem. However, participation in a depository or CNS system might virtually eliminate this problem.

*

From information obtained in discussions with personnel of the Security Pacific National Bank system.

-18

from 8 to 2 or 3 days. This would require a total investment of $50 million, or $5 million per year when amortized over a 10-year period.

Once this initial investment has been amortized, the unit cost per item transferred should be less than it is today. It is recognized that a move to machine-readable documents would probably facilitate automating the transfer process, perhaps resulting in a small investThis possibility, however, was not considered in making the above estimate.

ment.

Delivery Priority System. The primary requisite for operating an effective delivery priority system is approximately real-time information about the status of the firm's vault and delivery requirements. The ability to anticipate future delivery requirements must also be provided. Broker-dealers with automated cages (mostly large firms) already have this capability. It is recognized that the investment required of small firms to automate is prohibitive. They can, however, obtain the necessary information on a fee basis from service bureaus. The estimate of $8 million includes these fees. It is assumed that firms with automated cages would obtain the delivery priority capability in-house at a cost approximating service bureaus' charges. For a more complete discussion of delivery priority systems see pp. 24-26.

Partial Delivery System.

There are two major costs associated with operating a system of partial deliveries according to the existing partial delivery rule. First, more deliveries per trade are required. Second, there is the additional workload of making adjusting entries to keep books current that would otherwise remain in original form until closed out. The estimate of $6 million per year results from applying adjusted service bureau record keeping charges per delivery to the number of additional deliveries required. For further discussion see pp. 29-30.

Servicing Custodial Accounts. It has been suggested that the efficiency of the trade completion system would be improved if customers that typically take delivery of their certificates could be induced to leave them with their broker. The delay in receiving certificates from customers would effectively be reduced to zero, but at a cost. Custodial

-19

accounts must be serviced, statements prepared and mailed, and dividends and proxies collected and distributed. Also, the holding of certificates would increase the workload in the vault. At least one firm has quoted a charge of $40 per account for providing these services. The $40 charge seems high and probably was established to discourage customers from leaving their certificates with the firm. An estimate of $20 per account per year is considered adequate to defray the cost of providing the required separate services and is used in this study.

-20

III. ALTERNATIVE TRADE COMPLETION SYSTEMS

INTRODUCTION

In this section several possible changes in the trade completion system are analyzed. The alternatives considered constitute only a small sample of those available for consideration and should be regarded as suggestive rather than conclusive.

The general analytical approach used has been to establish one form of trade completion system as a benchmark and to compare its performance with a range of alternative systems. In this way, a constant reference point is provided and a systematic approach is always adopted. First, the effects of changes in each of selected variables are considered separately. For example, the first variable examined is that of alternative delivery priority schemes, the second, stock loan policies, and so on. Real choices are seldom available between such clear-cut alternatives; combinations are typically preferred. Since the objective is to find the optimum combination, the effect of various sets of changes is examined next. Delivery priorities are considered together with stock loans and partial deliveries, for example.

*

Each alternative trade completion system was specified by fixing the input values of the simulation model. The model was then run under a prescribed level of trading for 250 business days. The output from the simulation was the level of each kind of incomplete transaction generated. These were generated for each day and averaged over the 250-day period. The cost-estimating relationships discussed in Sec. II were applied to these results and a net cost was developed for each alternative. These costs provide the indexes for comparing the alternatives. In summary, for each alternative system, the level of fails is generated by the simulation model and the costs are estimated from the level of fails. The cost of fails indicates the relative performance of the alternatives.

*

The values used are the averages of the results of four runs using different random number series.

-21

THE BENCHMARK CASE FOR TRADING IN LISTED SECURITIES

*

The benchmark case was chosen to reflect the existing system as closely as possible, but under stress. Trade completion was simulated for securities traded on the NYSE at a level of 20 million shares per day, or about double the average mid-1970 volume. The model simulates trade completion for one security at a time, so the results were scaled to reflect the larger universe. Also, for simulation purposes, many general rules had to be specified, even though they were in conflict with some real-world conditions. For example, although it was observed that most firms make no serious attempt to follow a specific delivery priority, it appeared that deliveries are most often made to institutional customers first, to brokers second, for repayment of stock loans third, and to public cash customers last. This sequence, therefore, was specified for the benchmark case. Another example is the use of partial deliveries. Partial deliveries are sometimes made in the realworld, but they are exceptions and, so, were not included in the specifications of the existing system.

Based on empirical evidence collected in the spring and early sum+ mer of 1969, the uncompared trade rate was fixed at 6 percent, the bank DK rate at 30 percent, and the wrong denomination rate at 5 percent. The specifications adopted for the existing system may,

there

* "The scaling formula was: Total $ value for NYSE

=

(model result in share units) (100 shares per unit) (1500 issues traded per day) ($35 per share).

A full discussion of the empirical evidence collected for the study is included in Vol. III, ibid.