Lapas attēli
PDF
ePub

The Hypothesis About the Effects of Broker/Dealer Fiscal Management

To check the effects of broker/dealer fiscal management, we needed ind cators of fiscal pressures or financial health in our analysis. We included in ou analysis the information:

2 day of settlement before a trading day or not

25 buyer's net capital ratio

28 promptness with which buyer pays bills (a rating)

29 obligation buyer has in fails-to-deliver position (a rating)

42 settler's net capital ratio

45 promptness with which settler pays bills (a rating)

46 obligation settler has in fails-to-deliver position (a rating).

A high net capital ratio, indicating the broker/dealer has financial liabilities nearing the limit allowed by the 20:1 rule, or poor promptness in paying bills or : high obligation in his fails-to-deliver position each would indicate conditions of financial pressure for a broker-dealer. Payment for a settlement postponed over weekend is longer postponement, and therefore slightly more attractive when money is costing interest, than a postponement for 24 hours. We looked at the day of the business week on which settlement day occurred to see if this practice occurred as had been suggested to us. A relationship showing a high financial pressure to be associated with a high probability of a fail would confirm the hypothesis that broker-dealer fiscal management is an influence on fails.

The Dynamic Instability Hypothesis

The check on this hypothesis is an analysis of the effectiveness of the action which has been taken by the industry to correct the pyramiding of fails. That action has been fails clearing services offered by stock clearing houses in 1968 for the first time in the history of the securities industry. In principle, and in fact, it combines for a single member those certificates he owes in a particular security with those certificates he is owed and gives him a net balance position. These services reduce the nominal numbers of shares and numbers of dollars in a fail state at that point in time when the fails clearing is performed. If this process helps reduce pyramiding of fails, then securities for which a recent (sometime during the last two months) fails clearing has been performed should show a lower probability of failing than those securities for which the most recent fails clearing is more remote, or for which there has been no fails clearing. We included in our analysis the information:

49 time since most recent fails netting.

The Hypothesis About Market Volatility Effects

To assess the effects of price and price movement, we included in our analysis the following information:

3 price of stock at end of August 1968

4 variation in stock price in August 1968 (range/price).

This allows us to learn how price level for each security and recent volatility in the stock price has affected the probability of a fail. Another indicator-which we did not use in our study because of the high cost of getting the basic data-could be the amount and direction of price movement in the stock between trade date and settlement day.

CONTROLLING FOR OTHER CONDITIONS

Other conditions associated with a trade may mask the effects of the guessed causes, and we included number of kinds of information in our analysis to control statistically for these conditions.

Controlling for Broker/Dealer Size

Several indicators of broker/dealer size were used, and still others were available for use. We included in our analysis the information:

number of exchange/association memberships for buyer

22 number of buyer's employees

24 buyer's total assets

30 number of exchange/association memberships for settler
39 number of settler's employees (log)

41 settler's total assets

Controlling for Broker/Dealer Business Mix

Since both the mix of revenue producing work and the mode of doing the work varies from house to house, we attempted to take this into consideration by including the following information in our analysis:

19 buyer does underwriting or not

20 buyer clears for others or not

21 buyer clears through others or not

31 settler operates on agency basis with public or not

33 settler does underwriting or not

34 settler distributes mutual funds or not

35 settler makes third market in listed securities or not

36 settler clears for others or not

37 settler clears through others or not

38 settler has offices in NY, NJ, Pa., or not.

Controlling for the Size of the Settlement

Some trades and settlements are small, others large. Occasional comments from people in the industry suggested that the size of the settlement may even affect the probability of a fail. In any event, it seemed appropriate to control for the size of the settlement. We included in our analysis the following information:

[blocks in formation]

Since standards for listing on the three exchanges/associations are different, a rough statistical control for the differences in these standards was accomplished by including in our analysis the information:

7 security listed or not.

Controlling for Short-Term Variation in Workload

Trading volume fluctuates in the short term as well as the longer periods of time, and we controlled for the volume of trading on the three business days prior to the trade by calculating the combined trading volume in shares for the New York and American Stock Exchanges for the time immediately preceding each trade date. Our thought was that the increased or decreased work load in the operations offices could affect their performance and perhaps affect fails. We included the following information in our analysis:

1 trading volume for prior 3 days, NYSE and ASE.

ANALYSIS OF THE DATA

Samples with substantially complete data were chosen for analysis, and the relationship of each of these variables to all others was examined by calculating the correlations among the variables and examining these coefficients by factor analysis. The samples used in our analyses are described in Appendix B, and the method of analysis with some of its technical details is described in Appendix C.

V. THE CAUSES OF FAILS

In the previous chapter we showed how many different kinds of information were observed about each buy in our sample so that we could discover how they may affect the occurrence of a fail. We divided the buys about which we had complete information into two samples, and our analysis of the causes of fails is based on those samples.3 The relationship of each circumstance or variable affecting the trade to every other variable affecting the trade is represented by the correlation matrices presented in Table 6.

Each number in Table 6 is a correlation coefficient. Each coefficient indicates the degree of relation between two conditions affecting the buy.

For example, the buyer's number of employees and the buyer's assets are correlated +.57 in sample B and +.58 in sample C. These correlations, which are positive and much larger than .00, indicate that buyers with a large number of employees also have large dollar assets and a buying broker/ dealer with a small number of employees tends to have small dollar assets. The positive correlation between these two variables indicates that when one of the variables is large the other variable also is large. Since the correlation between dollar assets and number of employees in our sample of buying broker/ dealers is about +.58, we also know that these two variables are not perfectly correlated with each other. If that had been the case the coefficient would have been +1.00, a coefficient indicating the two variables are perfectly correlated with each other. Knowing the number of employees employed by a broker/dealer does not allow one to estimate precisely what the dollar assets of the broker/dealer are since these two variables are not perfectly correlated, but the relatively large positive correlation of +.58 indicates that broker/dealer firms of large size are characterized both by a large number of employees and large dollar assets.

As another example, it can be seen in Table 6 that the correlation between the event that the buyer and settler are in the same town and the number of miles between the buyer and settler is -.72 in sample B and -.71 in sample C. We have represented the fact that the buyer and settler are in the same town by the digit 1 and the fact that buyer and settler are in different towns by the digit 0. We have represented the distance

3. See Appendix B for a discussion of samples B and C.

« iepriekšējāTurpināt »