The bank requires us to keep a minimum balance; it is to our interest not to keep more than that. If we have more at the end of the day, we lend it out; if we have less, we borrow to make up the deficiency. We try to have just that balance, and no more, to our credit at the bank at the end of every day." The handling of funds by a brokerage house is a fine art, involving both technical skill and a philosophic grasp of the factors of the "money market." Are rates going up? Then it is well to reduce call loans, and borrow more on time. If lower rates are anticipated, more call money will be employed--with the possibility of a "squeeze" if too much is taken that way. Hidden dangers must be foreseen. The sums borrowed are enormous, and brokers' profits depend in very substantial degree on their skill in borrowing as cheaply as possible, and in utilizing their funds to the utmost. It is here, I think, in loan transactions between banks and brokers and between brokers, that we have a major part of the explanation of the huge deposit figures for New York City, and for the tremendous influence of stock sales on clearings, which Mr. Silberling's[445] figures show. This is the opinion of Professor O. M. W. Sprague, who first called my attention to the volume of call loans, and rapid shifting of call loans, in New York, and it is the opinion of every Wall Street man with whom I have discussed the matter. The actual pecuniary magnitude of the share sales and bond sales is not enough to do it. The mass of connected loan transactions, however, substantially greater in volume than the actual sales of securities, is, with the security sales, enough to do it. When the call rate is high, which will particularly happen when bank reserves are low, the shifting in loans will be much increased. One bank will have money to lend one day, but the next day will have to call it, to meet heavy demands at the Clearing House, while some other bank will have the surplus funds to lend. The brokers, by bidding up the rate, will tempt the temporary lending even of small surpluses, if their necessities are great. The volume of "all other deposits" and of bank clearings will be swelled by this much beyond ordinary. That this should not be revealed to ordinary statistical tests is due to the fact that speculation tends to fall off at such a time, so that the other factors in the stock exchange operations tend to reduce daily deposits and bank clearings. Mr. Silberling has applied to this problem the technique of a refinement of the correlation method, the method of partial correlation, with the result of confirming this view.[446] I conclude, therefore, that stock exchange transactions, instead of being undercounted in bank deposits, are very greatly overcounted.[447] The big item that does it is loan transactions between brokers and brokers and between brokers and banks.