-12

must put up capital to borrow the stock. The interest cost of the capital may reasonably be considered a cost to the borrower. On the other hand, the capital represents a benefit of equal amount to the lender. From the point of view of the brokerage community as a whole, the transaction nets out. Broker-to-broker fails also net out. These inter-broker transfer payments are specifically excluded from this analysis which is concerned with the securities industry as a whole.

Clerical and Administrative Costs

Clerical and administrative costs are reported under that heading in the 1968 Income and Expense Report of the NYSE. Analysis of these data for the 386 reporting firms indicates that of the $1.1 billion of clerical and administrative cost reported in 1968 approximately $180 million was associated with the failure of brokers and customers to deliver stock on settlement day. In this study, 4 percent of the average dollar value of broker fails-to-deliver to other brokers has been used as an estimate of the clerical and administrative cost of this kind of incomplete transaction. It is reasoned that this includes the cost of carrying the open item on the firm's books plus the costs of the research necessary to resolve the open items. A linear regression was performed with total clerical and administrative cost as the dependent variable and with average annual value of broker fails-todeliver to other brokers as one of the independent variables and total security commission income as the other. A regression coefficient for broker-to-broker fails of approximately .04, significant at the 95 percent confidence level, was obtained. To verify the estimator, a study was undertaken of the specific activities, relative to fails, that are carried on by a large NYSE member firm. The results indicated a figure somewhat higher than the 4 percent obtained from the Income and Expense Reports. However, because the 4 percent was based on a large sample, it was accepted and used here. Also, it seemed preferable to be conservative and understate rather than risk overstating the cost of fails.

There were no data available on similar costs associated with incomplete transactions between brokers and institutional customers, but

-13

it was reasoned that these also result in clerical and administrative cost. Brokers appear to make serious efforts to resolve failures to deliver to COD customers since these incomplete transactions usually tie up significant amounts of the firms' capital. It was assumed that 4 percent was also appropriate for the clerical and administrative costs associated with these fails.

Although fails to receive from all categories of customers and fails to deliver to public cash customers (prepaying customers) are of less concern to the industry, such fails nevertheless require that open items be carried on the books and that some effort be expended to resolve them. As is the case with fails to deliver to COD customers, there were no data available on fails involving deliveries from and to public cash customers. A cost of 2 percent of the average dollar value of this kind of outstanding fail has been assumed to reflect the existence of a cost, but one which is less than for the other types of fails. Stock lending and borrowing also incurs clerical and administrative costs. The borrower must find the lender. The lender must keep track of the loan, mark to the market periodically, and recall the loan at the appropriate time. Data obtained from the stock loan department of a large NYSE member firm showed that the clerical cost of each loan is approximately $5. This figure has been used to estimate the clerical cost of stock loans for the lender only; no cost is included for the borrower.

Interest Costs and Benefits

Certain incomplete transactions between brokers and their customers cause the level of capital in the hands of the brokerage community to be higher or lower than it would be if the transactions were completed on settlement date. When there is more capital, the industry benefits to the extent of the imputed interest on that capital. Conversely, when there is less capital, a cost is incurred at the same rate. Figure 1 shows the money flows between brokers and their customers that are related to stock transactions. Disruption of these flows causes interest costs and benefits.

Customers purchasing stock from brokers are of two types: those

that pay for their purchase on settlement day, irrespective of whether their stock is delivered to them, and those that pay only on delivery. Public cash customers constitute the first type. Failures to deliver to them have no impact on the level of capital and, hence, do not generate an interest cost. The second type consists, for the most part, of banks operating as agents for institutions or for their own trust accounts. Their dealings with brokers are on a COD basis, which means that no money flows into the brokerage community from a stock purchase until the stock is delivered. It follows that a failure on the part of a broker to complete delivery of stock to a bank delays the receipt of capital and results in an imputed interest cost.

When customers sell stock, they receive the proceeds of the sale from the broker when they deliver the stock sold. To the extent that delivery of the stock is delayed, capital is retained in the system, and the system benefits in the amount of the imputed interest on the capital retained. As all transactions between brokers and customers (when customers sell) are on a COD basis, all fails-to-receive from customers result in an imputed interest benefit. In this study, an interest rate of 9 percent has been used. These factors for estimating the annual cost associated with such fails or incomplete transactions are as follows:

Cost of Trade Completion System Improvements

The cost of these improvements is the third type of cost considered. Although design studies and detailed cost estimates have not been made, the estimates of system improvement annual cost are believed to be consistent and sufficiently accurate for use in an initial screening of a wide range of alternative trade completion systems that was a goal of this study. The costs of improvements listed below are explained in detail on the following pages:

Certain of these estimates, such as for delivery priorities and for partial deliveries, depend on the level of market activity. In these cases, a level of trade-completion activity generated by a NYSE trading volume of 20 million shares per day has been assumed. The reasons for making this particular assumption will be discussed later.

-16

Bank DKs. As explained earlier, the DK occurs when a broker attempts to deliver securities to a bank, and the bank refuses to accept the stock, claiming no knowledge of the transaction. The bank says,

"I don't know the trade." There are numerous reasons for this. An important one, and the one dealt with in this analysis, is that the bank, acting as agent, does not have an authorization from its customer to accept the stock delivery and make the payment. Data from a large brokerage firm and from representative banks indicate that about three DKs result from every ten attempted deliveries to banks and that, of those three, 60 percent are attributable to a lack of instructions. This was observed to be largely a problem of communication between banks and between banks and their clients. The cost of eliminating this type DK is estimated to be $6 million per year. This is the estimated cost of having the appropriate party send instructions, by wire, to the receiving bank before settlement day. Estimates of the number of instructions required were made and translated into cost, using rates associated with available Western Union services. Leased lines or other bank-controlled private systems are certainly possible and would probably cost less. The $6 million is, for that reason, a conservative estimate. Doing away with DKs that result from lack of instructions would reduce the DK rate from 3 in 10 to 1 in 10, or by two-thirds.

In

Wrong Denomination. A wrong denomination problem exists when a broker-dealer firm has sufficient stock to make a delivery, but the certificates available are of the wrong size. For example, a firm needs to deliver 100 shares but has only a 500-share certificate. this case, the 500-share certificate would have to be sent off to a transfer agent to make change before the delivery could be accomplished. The larger firms, those with sizeable inventories, should be able to reduce the frequency of this problem by some form of statistical inventory management. Past deliveries could be examined and requirements for future deliveries anticipated. In this way, a supply of certificates in appropriate denominations could be maintained. It is estimated

*

*For small firms, those that seldom have inventories of any appreciable size, this would not be feasible. There seems to be no economical