The Stock Market, Credit and Capital Formation

APPENDIX B

STATISTICAL NARRATIVE FOR THE UNITED STATES

How large were the funds which flowed onto the securities markets in a certain period? What was the origin of the funds which were devoted to the purchase of securities? What did the brokers do with the funds received; to whom were they paid out? What did the sellers of securities do with their sales proceeds? How large were the funds which came to corporations for newly issued securities, and what did the corporations do with the funds received? How much of these funds went into real investment?

These and other questions ought to be answered in a statistical narrative. But it cannot be done. No information is available that would enable us to get even near a satisfactory answer. The statistical data we do have contain far less information than has often been believed. Naïve interpretations have led to conclusions which prove untenable on closer inspection.

The only statistics which can be produced to show certain “funds connected with security transactions” are the statistics on brokers’ loans. It is, of course, a purely arbitrary procedure to attach much significance to these data because the funds actually employed for security purchases may be a fraction or a multiple of the figure of brokers’ loans. We have no figure of the amount of their own funds which security buyers applied to security purchases; we have no figure of the funds which they borrowed through others than brokers. If we had the figure for the amount of their own funds that were put to security purchases, we should still not know very much, unless we knew how much of these funds were proceeds from previous sales of securities, how much were proceeds from previous sales of other kinds of assets, and how much were new voluntary savings from income received.

As to the funds which security buyers borrow from the banks directly, i.e., not through their brokers, one might choose to take the statistics of security loans of banks. However, these loans on securities as collateral need not be used by the borrowers for purchases of securities. It is generally known that a large part of collateral loans are made to commercial borrowers for business purposes. But what part? There is no use guessing. There is no use assuming a constant proportion between collateral loans for business purposes, for consumption purposes, and for speculative purposes, because the proportion is undoubtedly shifting in the course of time, both secularly and cyclically. As I am ignorant of the portion that is used for security purchases, I refrain from reproducing here the statistics of the security loans of banks.

As the one series of figures whose connexion with the stock exchange cannot be doubted, the statistics of brokers’ borrowings from 1926 to 1937 are shown in Table I. This table contains the borrowings of New York Stock Exchange members only, but the borrowings of brokers on the minor stock exchanges in the United States would not increase the figures very considerably.

It was the phenomenal rise of the New York brokers’ borrowings from 3219 million dollars in September, 1926, to 8549 million dollars in September, 1929, which brought the subject of stock-exchange credit into the limelight. It was these statistical series which were believed to inform us about the alarming “absorption” of credit by the stock exchange.

The data on brokers’ borrowings are reported by the New York Stock Exchange. Separate figures are given for borrowings from New York City banks and trust companies, and for borrowings from other sources. These other sources include other brokers. What amount of “inter-broker-loans” are among the borrowings from others than New York City banks is unknown; the New York Stock Exchange has information from its members on the “total of money loaned in firm name exclusive of money loaned as agent of others” only from 1932 onwards. Assuming that this money was loaned by brokers (directly or indirectly) to other brokers, the figures published here (with the kind permission of the New York Stock Exchange) in Table II may be taken as representative of the volume of inter-broker-loans from 1932 to 1935. These figures are relatively small; the money loaned out by brokers was (in September, 1935) as little as .44% and (in August, 1932) no more than 6.78% of the brokers’ borrowings. Whether inter-broker-loans were an equally insignificant (or still less significant) part of brokers’ borrowings during the boom years cannot be ascertained, though it seems probable from deductive reasoning.

Table I does not tell whether the brokers’ borrowings-from the New York City banks were loans on these banks’ own or on others’ account. This information is given in the statistics of brokers’ loans of New York City banks, published by the Federal Reserve Board. Because of a change in classification since 1926 these statistics are reproduced here in two tables; Table III showing Street Loans of New York City Banks, 1922-1925, and Tables IV and IVa showing Brokers’ Loans of New York City Banks, 1926-1935 and 1936-1939 respectively.

The totals in Table IV will be found to conform approximately to the figures in the corresponding column of Table I; “the slight differences are due not only to the different dates of compilation, but also to the fact that Reserve members sometimes lend to stock brokers and security dealers not members of the exchange, and also stock exchange firms sometimes borrow from New York banks and trust companies not members of the Reserve.”1

Until 1935 New York City banks made three classes of loans to brokers: loans on account of banks outside of New York City, i.e., they acted as agents for out-of-town banks in placing funds of the latter on the New York call money market; loans on account of others, i.e., they acted as agents for non-banking firms in placing funds on call; and finally loans on their own account. In Chapter VII of this book, reasons were advanced as to why loans “on account of others” are of a totally different nature than genuine bank loans. It appears that these loans on account of others either liquidate existing bank loans (and thereby release bank reserves) or represent no funds at all (being merely the statistical expression of the fact that sellers of securities loan to buyers of securities, with no funds involved in the transaction). For the problem of the flow of funds to or through the securities market or for the problem of “credit absorption,” it is not the total of brokers’ loans by banks, but only the brokers’ loans by banks for their own account or for account of other banks which are really relevant. Compared with the phenomenal rise which the total of brokers’ borrowings, and the total of all bank loans to brokers, showed during the latter part of the twenties, the rise in brokers’ loans for account of banks (i.e., exclusive of the loans on account of others) is less alarming. Between September, 1926, and September, 1929, brokers’ loans for account of out-of-town banks rose from 1128 to 1850 million dollars, and brokers’ loans for the New York member banks’ own account rose from 974 to 1048 million dollars; hence together they rose from 2102 to 2898 million dollars.

The brokers’ loans on account of others fell rapidly after the crash of October, 1929. They fell by more than 50% within two months, and then declined steadily. In 1931 they amounted only to about 5% of the record volume; they were further reduced by a change in clearing house rules in November, 1931, and later outlawed by the Banking Act of 1933. The last million disappeared in May, 1935.

The statistics of Tables I to IV refer only to New York brokers or New York City banks respectively. Figures for the whole country are available in the statistics of “All Member Banks.” This information, in contrast to that on the “weekly reporting member banks in 101 leading cities,” is confined to three or four call dates every year. The statistics of brokers’ loans by all member banks from 1928 to 1939 is given in Table V. In order to allow a comparison of the volume of brokers’ loans with the total of all loans, and with the total of all loans and investments by member banks, the respective figures are reproduced in Table V, together with the calculated percentages.

The comparison shows that the brokers’ loans by member banks amounted to 14.04% of all loans in December, 1928, while they were lower than that during 1929: the absolute fall of brokers’ loans by banks (on their own account) and the rise of other bank loans during 1929 reduced the percentage to 10.79 on October 4, 1929 (before the crash). The percentage was higher again in 1930, and began to fall only in 1931, reaching a low of 3.38 in June, 1932. It was partly due to the fall in other bank loans, but mainly to a recovery of brokers’ loans, that the ratio of brokers’ loans to all loans rose again and exceeded 10% in June, 1934, and from December, 1935, to June, 1937.

The general change in the composition of bank assets, in particular the steady increase in security holdings of banks from June, 1933, to December, 1936, made the ratio of brokers’ loans by all member banks to their total loans and investments behave differently. The ratio of 9.90% in December, 1928, has not been approached at any time since. From September, 1931, to June, 1938, the ratio of brokers’ loans to all member bank loans and investments has never reached 5%.

Attention should be drawn once again, however, to the fact that brokers’ loans are by no means the only way in which bank funds may be used for stock-exchange purchases. We have already referred to the security loans by banks, and might mention here that during the boom period of the twenties, security loans comprised up to 30% of the banks’ total loans and investments.

In connexion with the statistics of brokers’ loans it is appropriate to reproduce the statistics of the rates charged on stock exchange call loans. Table VI furnishes these figures. We can see that the monthly average of call rates rose from 2% per annum in August, 1924—the low of the twenties—to 9.8% in March, 1929. The low rate of 2% was reached again no later than in 1930. A new low was recorded in 1935 when the rate stayed for five months at .25%. From June, 1936, on, the rate was absolutely stable at 1.00%: competition has apparently been restricted on the call money market.

One of the ways in which funds may flow from the stock exchange into industrial circulation is through new security issues. Just as brokers’ loans were not the only type of inflow to the stock exchange, capital issues are not the only type of outflow. Sellers of old securities can withdraw their proceeds and use them for purchases of products and services just as sellers of new securities can. Real investment can be undertaken out of sales proceeds from old securities just as out of sales proceeds from new securities. The statistics of capital issues may thus show figures which may represent only a fraction of the funds actually put to new real investment; but they may equally well be a multiple of the funds actually so employed, because the capital raised may not have been used for investment purposes at all. Apart from refunding operations, the funds raised by corporations may have been used for purchasing assets from holders who may use the proceeds for absolutely anything—including consumption, security purchases, lending out, accumulating idle balances or what not. Or the corporations themselves may use the funds for all these things, including call loans to the stock exchange which indirectly involve loans to the buyers of their securities; in this case no funds flow either to the stock market or to the issuing corporations (see Chapter VII).

With this proviso we reproduce the statistics on Capital Issues. Table VII gives annual figures from 1922-39 of all flotations of bonds and stocks, new and refunding, domestic and foreign, government and corporate. The rise and fall of the total issues from 5119 million dollars in 1923, to 11,513 million dollars in 1929, and to 1063 million dollars in 1933 is impressive enough; yet it is less impressive than the rise and fall in corporate stock issues from 570 million dollars in 1922, to 5924 million dollars in 1929, and to 20 million dollars in 1932.

These securities were issues by industrial and by financial corporations. There is no doubt that the proceeds from the issues of financial corporations did not flow directly into industrial circulation. We do not mean to imply that the proceeds of the issues of industrial and other non-financial corporations did necessarily flow into industrial circulation; we have said expressly that they did so only in part (and perhaps only in small part); the whole of the capital raised by financial corporations certainly remained in the financial circulation for further transactions. For this reason it is appropriate to separate the issues of financial and of non-financial corporations as is done, beginning from 1926, in the series of Table VIII.

In this Table VIII the monthly figures of new security issues by corporations are shown. It is interesting to see that the new stock issues by financial corporations which usually amounted to only a small percentage of the stock issues of non-financial corporations (4.94% in 1926, 6.30% in 1927) rose enormously in 1928 and 1929 (26.76% in 1928 and 52.84% in 1929) and indeed exceeded the non-financial issues in August and September, 1929.

It is a commonplace to state that corporations are likely to issue new shares when they can sell them easily and at good prices. Nevertheless, it may be useful to have the series of corporate stock issues and the series of stock prices side by side with the figures of stock issues freed from the fortuitousness of the calendar month by means of three-month moving averages. Table IX presents these series from 1922 to 1939. Three-months moving averages of the new stock issues of non-financial corporations are shown with the stock price index; the result is the expected one: the amount of capital raised generally varies directly with stock prices.

A chart showing these series in two curves might serve to demonstrate the speed (or absence of lag) with which corporate capital issues react on higher stock prices. This reaction might be expressed also in another form. The corporations offer new shares against money capital, i.e., they demand money capital for which they pay in new shares. If stock prices are low, corporations have to give many shares for 100 dollars of money capital and they will not demand large amounts; if stock prices are high, corporations have to give fewer shares for 100 dollars of money capital and they will demand larger amounts. The reciprocal of stock prices can thus be considered as the price of money capital in terms of corporate shares. If stock price movements were the only changes in data, the relative change in money capital raised might be expressed by the elasticity of demand for money capital in terms of corporate shares.

It can be argued that the demand for money capital in terms of shares is extremely elastic, but this cannot be proved by statistics. If the amounts of money capital raised are plotted along the abscissa, and the reciprocals of the stock price index along the ordinate, the scattered points in fact suggest the form of an elastic demand curve—but to accept this as a proof is to overlook the fact that in times of low prices of money capital (i.e., in times of high stock prices) profit expectations of firms may be increased so that the cheap and easy money is by no means the sole attraction for issuing corporations.

The figures on foreign capital issues included in Table VII (as well as the considerations in Chapter X) invite inspection of the statistics of capital movements (both long and short term) into and out of the United States. Table X reproduces the figures on capital movements as contained in the United States Balance of International Payments, 1922-1938. The net movements of long-term capital do not seem to permit of an interpretation in terms of cycle analysis or stock market analysis. Whereas net outflows of long-term capital were observed from 1924 to 1930, net inflows were the order from 1931 on. The explanation of this circumstance is likely to be in the political field. Prosperity and depression, security market boom and slump, seem to affect not the direction but merely the volume of capital movements in both directions. Both inflow and outflow of long-term capital increased considerably from 1924 to 1928, decreased from 1929 to 1932, and increased again from 1934 to 1936. The excess of outflow above inflow, or inflow above outflow, does not reveal any cyclical influences.

The net movements of short-term capital also resist interpretation in terms of cycle or stock market analysis. These movements of short-term balances due from and to foreign countries are the resultants of so many complex forces that a single-track relationship, or a correlation with any one factor, could not be expected.

None of the statistics discussed so far has given us any clue or hint as to the problem of “absorption of funds” by security markets. Three tests would be necessary to prove or disprove the “absorption” hypothesis. As stock exchange credits rise (1) does or does not the increment of these loans or a considerable part of it appear in an increase in cash balances held by brokers? (2) Does or does it not appear in an increase in idle cash balances held by the sellers of securities? (3) Does or does it not appear in an increase in cash balances of private speculators and financiers (individuals and firms) who switch the funds back and forth among themselves in repeated financial transactions?

No statistical information whatsoever was available during the twenties which would have permitted an answer to the three questions. Some reliable estimates can be said to have answered the first question in the negative: the brokers’ cash balances were found to be so trivial in relation to the brokers’ borrowings that to allege credit absorption in this form would have been almost ridiculous. One of these estimates hit upon an amount of about 20 million dollars as the average of New York brokers’ balances from 1922 to 1926, i.e., at a time when the New York brokers’ borrowings amounted to between one and three and a half billion dollars.2 According to other findings the ratio of “cash in bank” to “total borrowings” was 2.28% in a representative case, checked by a number of other tests.3 The same source estimated brokers’ balances in September, 1929, at about 100 million dollars as compared with the 8549 million dollars borrowed by brokers.4

No material was or is available to allow estimates or guesses concerning a possible accumulation of idle balances by security sellers, and concerning the operating balances on bank accounts of speculators and financiers. Attempts have been made to find some clues by studying changes in the geographical distribution of deposits and reserves.5 It was found that the rise in New York brokers’ borrowings in some periods was, and in other periods was not, accompanied by a redistribution of bank deposits and reserves between the New York area and other regions. These findings are not particularly indicative in either direction. Bank deposits may be kept idle or kept waiting for further financial transactions by provincial holders no less than by New York holders.

Information about brokers’ cash balances has been collected and published for recent years (from September, 1935, on) by the Federal Reserve Board. These figures are reproduced in Table XI. The cash balances carried by brokers are higher than had been estimated for the twenties: they fluctuate between 179 and 268 million dollars as compared with the 20 and 100 million dollars respectively in the estimates for the early and late twenties. Were these estimates too low or do the brokers now carry much more cash than some 10 years ago? Probably both are true to some extent. Among the several reasons which might explain why brokers now hold higher bank balances than in the past, are the stricter rules imposed by supervising authorities, and the low money rates. The latter make the cost of carrying cash lower than the inconvenience of repaying and rearranging call loans.

Besides brokers’ cash balances, Table XI shows other ledger balances of New York Stock Exchange members: their margin loans to customers (“customers’ debit balances”), their total borrowings (“money borrowed”), and the customers’ deposits with them (“customers’ credit balances”). These series cover only a relatively short period (September, 1935, to July, 1939), but a period which contains a conspicuous rise, a break and a sharp fall of stock market values. Before any comparisons between the various series are made it should, however, be emphasized that the absolute and relative magnitudes during this period were very different from those of the boom period in the twenties. In the period covered in Table XI, brokers’ borrowings rarely exceeded one billion dollars, a very modest figure compared with the eight and a half billion dollar record of 1929. Customers’ deposits rarely exceeded 400 million dollars, whereas they undoubtedly reached a multiple of this figure in the late twenties. With brokers’ cash balances higher and all the other items lower than in the twenties, the ratios between the various ledger balances are, of course, totally different now from what they were then.

The percentage ratios of the cash balances to each of the other items are given also in Table XI. The way in which the percentages fluctuate strongly suggests absence of correlations. The ratio of cash to margin loans to customers varied from 13.39% to 27.78%. The low was in August, 1937—when the margin loans were the third highest of the period—the high was in June, 1938—when margin loans were the second lowest.

The ratio of brokers’ cash to their borrowings varied from 16.38% to 43.43%. Again, the low was in August, 1937, when borrowings were highest, the high in June, 1938, when borrowings were next to the lowest. The poor hypothesis that brokers’ borrowings rise because of the brokers’ increased demand for cash balances would have needed the reverse of these results for its support.

The comparison of our ratios of cash to borrowings of between 16.38% and 43.43% with the ratio of 2.28% estimated for autumn, 1929,6 is striking. Even if the cash balances in 1929 had been considerably higher than they were estimated to be, the ratio of cash to borrowings would still have been extremely low. It seems thus more than obvious that the considerable changes in brokers’ demand for cash balances are neither cause nor effect of changes in the brokers’ borrowings.

The third ratio shown in Table XI is that of brokers’ cash balances to customers’ brokerage deposits. There are again wide fluctuations of this ratio: between 46.17% and 69.39%. The range of fluctuations seems relatively smaller than in the two other ratios, but this is only due to the arithmetic of the matter; in the two other cases the denominators of the fractions (loans to customers and brokers’ borrowings respectively) were much larger than in the third case (customers’ deposits). The ratio of brokers’ cash to customers’ deposits is so surprisingly high, not because brokers are in the habit of holding so high a “reserve ratio,” but rather because brokers’ cash balances are not dependent on customers’ deposits and cannot be reduced when customers’ deposits fall. Likewise, the cash balances need not be increased when customers’ deposits rise. One may thus assume that the ratios of cash to deposits were incomparably lower in the boom period of the twenties, when customers’ deposits were undoubtedly very high. Unfortunately no figures are available for that period. It seems, however, obvious that customers’ balances with brokers must be higher when many customers are engaged in continuous in-and-out trading. Even if customers reinvest immediately after each sale and have their credit balance only for a short interval, the sum total of trading accounts must show a high credit balance in such periods.

The hypothesis that brokers’ cash balances must rise when stock prices rise and/or when brokers’ borrowings rise does not hold water. This has been shown by deductive reasoning. The figures which are available from September, 1935, on might be considered further evidence if such were needed and if so short a time series were acceptable as evidence. Table XII facilitates a comparison of brokers’ cash balances with stock prices and brokers’ borrowings by giving the index figures of the three series. January, 1936, has been chosen as the base, because the Federal Reserve index of common stock prices, which uses 1926 as a base, happens to be at 100.1 in January, 1936. To take this as our base after eliminating the decimals saved us recalculating this index series. For the same reason the Federal Reserve index was used here in contrast to Table IX where the Standard Statistics stock price index was used.

The indices of brokers’ cash balances and of brokers’ borrowings were calculated from the original figures contained in Table XI. The peak of the cash index was 139 in April, 1936, when the stock price index was only 109 (as against its peak of 130 in March, 1937) and the index of borrowings was only 112 (against its peak of 134 in August, 1937). Another high of the cash index was 137 in October, 1937, when both the index of stock prices and the index of borrowings were low (91 and 85 respectively) and, moreover, falling. The coefficient of concurrent deviations from September, 1935, to July, 1939, was negative (r = – .36) for brokers’ cash balances and stock prices, and was equally negative for brokers’ cash balances and brokers’ borrowings. Although it is not intended to draw conclusions from the scant statistical material at our disposal, the figures presented strongly suggest the opinion that brokers did not borrow in order to hold cash balances, nor did they hold cash balances because of high stock prices.

The belief that brokers must hold higher cash balances when the volume of transactions rises, prompts a comparison of the relevant figures. Since, however, the volume of payments arising out of stock transactions is not proportional to the volume of stock transactions, the relation between transactions and settlement has to be examined first.

The New York Stock Clearing Corporation publishes statistics on the “obviation of the use of bank credit,” i.e., on the absolute and relative amounts of transactions which were settled through the clearing mechanism and, thus, did not give rise to cheque payments. The annual averages of “percentage obviation” are reproduced in Table XIII which gives the theoretical obviation of cheque payments from 1925 to 1938. Monthly figures, available only from March, 1929, on, are reproduced in Table XIV, which gives total clearing house transactions and clearing house balances settled by cheque together with the percentage obviation. The highest obviation figure during this period was 85.4% (in August, 1932), the lowest 61.1% (in September, 1934). The former refers to cheque payments of 496 million dollars settling 3385 million dollar transactions; the latter refers to cheque payments of 464 million dollars settling 1195 million dollar transactions. These figures illustrate our thesis that a higher volume of security transactions need not give rise to a higher volume of payments.

Closer inspection of the series will, however, show that the “funds actually required,” i.e., the volume of cheque payments, did not remain entirely unchanged when the volume of transactions increased. The percentage of payments to total transactions decreased, of course, when transactions rose, and increased when transactions fell; but the absolute amount of payments still, though only slightly, increased with higher transactions, and decreased with lower transactions. The highest volume of payments coincided with the highest volume of security transactions (October, 1929); and, likewise, the second lowest volume of payments coincided with the third volume of security transactions (February, 1938). The explanation of the concurrent movements seems to lie in the fact that increased business frequently involves a somewhat uneven distribution of the increased selling and buying orders between different brokers.

Do increased security transactions, which may thus be accompanied by increased clearing balances to be settled by cheque, cause brokers to carry larger cash balances? With the statistics of brokers’ cash balances available since September, 1935, the series can be compared for the last four years. Table XV is concerned with the volume of transactions on the stock exchange, with the actual cheque payments settling these transactions, and with the brokers’ bank balances. Many of the calculated figures of this table suffer, however, from an irreparable defect. While transactions and settlements refer to a period of time (months), cash balances refer to a point of time (end of months). Similar calculations (e.g., of velocities of circulation) usually adopt the method of comparing an average of daily cash balances with the amounts transacted or settled during the period. Such a procedure was not possible in our case because the end-of-the-month figures of cash balances are the only ones available. To take an average between the cash balances of the end of the month and those of the end of the preceding month seemed too arbitrary a makeshift.

The total value of clearing house transactions was again taken as representative of the volume of security market transactions, although we should find that there are differences between the figures reported by the clearing house and the figures reported for the total value of securities sold. The differences are due to delayed deliveries and settlements, and to securities loaned via the Stock Clearing Corporation.7 For a comparison with the actual settlements by bank cheques the value of clearing house transactions is certainly the appropriate figure to take.

During the period covered by Table XV the highest volume of clearing balances settled by cheque was 785 million dollars, in January, 1936; at the end of that month brokers’ cash balances amounted to 193 million dollars. A very low volume of clearing balances settled by cheque was 269 million dollars, in February, 1938; the brokers’ cash balances carried at the end of that month were 207 million dollars. These figures suggest anything but a correlation. The cash balances in March, 1936, were relatively the lowest, with 23.12% of the payments of the month and 5.34% of total transactions; the cash balances in February, 1938, were relatively the highest, with 76.95% of the payments of the month and 20.84% of total transactions. It seems fairly obvious that the changes in the relative cash position are chiefly the result of fluctuating transaction volumes and insensitive cash balances. With a coefficient of concurrent deviations from September, 1935, to June, 1939, of only + .49, the assumption of a correlation between brokers’ cash balances and security transactions does not seem to be warranted.

If there is no, or only little, correlation between cash balances and transaction volumes, it is obvious that there must be a fairly high correlation between transaction volumes and the velocity of circulation of the cash balances. Velocity of circulation of the cash balances may be understood as the ratio to cash balances either of total transactions or of actual payments. The former would be “virtual velocity” comprising the efficiency of the clearing mechanism, the latter would neglect the transactions that were offset through the clearing procedure. Since we are used to expressing velocity as the number of times a thing is turned over per year, the monthly figures of Table XV were, for the computation of velocities per annum, multiplied by 12. The lowest velocities during the period covered were 15.6 with reference to actual cheque payments, and 55.3 with reference to total transactions. The highest velocities were 51.9 and 224.8 respectively.

It should be noted that neither of the velocity figures that refer to brokers’ cash balances are, in point of fact, descriptive of the stock exchange settlement mechanism, for the following reason. For the payment of clearing balances, brokers employ “bank funds” which are on their accounts neither at the beginning nor at the close of the business day. The broker who has to pay an adverse clearing balance usually arranges a bank loan, in the course of the day, while the broker who receives payment for a favourable clearing balance usually repays a bank loan before the close of business. No credit balances remain on the brokers’ bank accounts from these transactions. Where brokers borrow or use their favourable clearing balances for payments to customers, it is again not the brokers who carry a cash balance.

Thus, bank funds may be said to be employed even if they do not show in the balances of the close of the day. If one bank lends and another is repaid, the latter cannot usually relend before the next day. Similarly, if one bank lends and brokers’ customers obtain “derived” deposits with other banks, it is usually only on the next day that these deposits can be drawn upon. For this reason all new brokers’ loans were said to be “absorbed” for one day. On the basis of this notion a velocity of circulation of “security-trading funds” was calculated. These “security-trading funds” included brokers’ cash balances (at the close of the day) plus new brokers’ loans (placed during the day) plus new brokers’ deposits of receipts (deposited during the day but possibly offset by withdrawals).8

The “security-trading funds” in the sense discussed above are bank funds. The trading funds of the public, however, may be, in large measure, brokerage deposits, i.e., customers’ free credit balances with brokers. It is possible that a greatly increased volume of security transactions may be completed with an increase in velocity of circulation of these brokerage deposits without any increased payments of clearing balances between brokers, provided selling and buying orders are more evenly distributed among brokers. It is also possible that a greatly increased volume of securities may be bought with a great increase in customers’ margin debts without any increase in payments between brokers and without any increase in brokers’ loans and even without any shift of loans between brokers, provided that margin buying and selling with the intention of further trading are evenly distributed among brokers. In such a case the increase in customers’ debit balances would create an increase in customers’ credit balances, and no bank funds whatsoever would be involved in these transactions.

No statistical information is available to show the extent to which the possibilities described were actually realized. The statistics of transactions and customers’ balances, given in Table XVI, cannot throw much light on this question. The customers’ balances9 are end-of-the-month figures, whereas only average balances could be legitimately compared with the monthly transaction volumes. The correlation between the three series is not high enough to be useful as evidence either way. It is noteworthy, however, that of the 21 times that, between January, 1932, and June, 1939, the customers’ credit balances exceeded 300 million dollars, 14 times were in months during which security transactions had exceeded two and a half billion dollars. If it is true that no increase in brokers’ cash balances need accompany increased transaction volumes, the idea of “self-generated” security-trading media (in the form of brokerage deposits) suggests itself once more. Unfortunately no information on these matters is available for the boom years of the twenties.

_________________

With the scant material available, the statistical narrative of stock market behaviour in the United States had of necessity to be meagre. Nothing much more than a few sidelights on the relevant questions could be obtained from the examination of the data at our disposal. The more significant of our statistical series did not go back far enough to permit sufficient verification of the findings of our largely deductive analysis. This is regrettable because it is the past “heroic era” of the stock exchange for which factual information would be especially important.

It would still be possible, I think, to uncover more information about the years of greatest stock market activity. Records and books must be available in brokerage offices and in banks. The movements of brokers’ bank deposits and of customers’ brokerage deposits (and the turnover of these accounts) might be obtained for a representative sample. These series would have to be compared with the corresponding series of stock market transactions, stock prices and brokers’ loans. Most revealing would be weekly or monthly statements of the brokers’ “balance of payments,” with the payments from and to customers broken down with respect to various types of customers such as individual traders, industrial corporations, financial corporations, &c. While this information may be obtainable with no over-serious difficulties another piece of information, perhaps the most crucial of all, seems to be inaccessible: the uses to which customers put the balances which they withdraw from their brokers. An inquiry extending to the brokers’ customers would appear to be impracticable.

It may be worth considering whether something like the above-mentioned “balance of payments” statements of broker firms might not be obtained currently. To report at least the figures of the (gross) payments received from customers and (gross) payments made to customers during each month may not be too heavy an additional burden on brokers’ offices (although they are, it is true, already badly inconvenienced by their existing current report duties). To know the total gross amount of payments from brokers to customers (and the ratio of this amount to total transactions) would be a valuable supplement to our knowledge, from the Stock Clearing Corporation statistics, of the total amount of inter-broker payments (and the ratio of this amount to the total transactions).

There are other pieces of information which one might obtain from more detailed bank statistics. We know the amounts of brokers’ loans outstanding at certain dates, but it may be interesting also to know the gross movements of brokers’ loans. The brokers’ loans outstanding are, for the reporting member banks, end-of-the-month figures. The daily gross movements of brokers’ loans, inclusive of the so-called day loans (which are arranged and repaid before the close of business of the day) might tell us more about the current procedures. The gross amount of brokers’ loans granted during a day may easily be a multiple of the net increase in loans outstanding at the close of the day. This may have implications with regard to the utilization of the banks’ reserves, on the one hand, and the “virtual” velocity of circulation of the brokers’ bank funds, on the other.

Thus we come back again to the problem of the financial circulation, and we may wonder whether it would not be possible to have the statistics of bank deposits and bank debits broken up according to types of depositors such as financial firms, industrial firms, large individual depositors and small individual depositors. Such a decomposition of the figures representing the national money flow would certainly be most enlightening. The expense may be higher than could be imposed upon the banks without compensation, but the information obtained may be well worth the cost.

________________

Rogers’ method was employed recently in a slightly modified form by William M. Blaisdell, Financing Security Trading (1935). The results are not very enlightening. Blaisdell assumes that the new brokers’ loans are used once per day, hence 250 times per year if only stock market days are counted. (Blaisdell deducts 10 per cent, in order to account for inactive service balances; however, compensating balances or service balances are seldom required for brokers’ loans.) This (assumed) velocity figure is multipled by a coefficient of clearing efficiency in order to account for check obviation. If the obviation is 80 per cent, total transactions are 5 times the check payments; virtual “velocity” of these funds is then 1250. According to the fluctuations in the obviation rate Blaisdell’s velocity figures vary between 1000 and 1600.