-27

to find the funds for loan growth. Accordingly non-member banks merely would appear to act less procyclically than member banks during periods of monetary restraint. Now consider recession. Member banks, having so sharply reduced investment growth in the boom, now feel a need for a much higher degree of safety and less risk in their asset portfolios. Accordingly their loan growth would drop sharply as their investments

rise.

Non-member banks, not having been so severely pressed for funds in the boom, already have a greater fraction of their assets in relatively safe form and therefore do not need to add investments to their portfolios as rapidly as member banks. The net effect is that non-member bank investments may grow more slowly in recession than those of member banks.

Preliminary results of a statistical model of loan and investment behavior. Given the difficulties of interpreting the bank loan and investment behavior of non-member and member banks, I have made a preliminary effort to set up a statistical model of how bank loans and investments respond to various macroeconomic variables. Since the study is only in its initial stage, I shall only briefly outline the procedures used and the tentative results.

The dependent variables examined are: (1) non-member bank loans and investments, LI ; (2) member bank loans and investments, LI ;

nm

m

(3) country bank loans and investments LI; (4) all commercial bank loans and investments, LI ; (5) non-member bank loans, L ; (6) member

ac

nm

bank loans, L; (7) country bank loans, L; and (8) all commercial bank

loans, Lac

Each of these dependent variables is regressed against

macroeconomic variables in a multiple regression equation.

0

-28

The macroeconomic variables are: (1) i, the prime interest rate;

(2) Y, personal income; (3) P, the consumer's price index; and (4) the unemployment rate, U. It was expected that LI and L would be positively related to i, Y, and P, but inversely related to U. Since commercial banks trade off loans and investments, it was expected that the relationship between LI and the independent variable would not be as good as that between L and those variables. Bank reserves were not included in the model as a variable due to the difference in the meaning of reserves for member versus non-member banks. The period studied covers 1957 through 1970. The regressions were performed using first the absolute figures and then first differences.

as is

Income

2 The data for total loans and investments shows a high R to be expected when using time series and absolute figures. was the most significant variable. As noted earlier, it was expected that LI would vary positively with Y, i, and P and negatively with U. The expectations for Y and U were confirmed by the test in absolute Overall the results seem best for non-member banks, thereby giving some credance to the conjecture that non-member banks react more procyclically than do member banks.

values.

The results concerning loans show a somewhat different pattern for non-member banks. L are positively related to all the independent

nm

variable, indicating that, aside from changed sign relative to U,

non-member banks behave even more procyclically than do their total

loans and investments.

The tests in first differences were designed to eliminate serial

correlation.

Of course this was at the expense of the high R2 and the

ว

-29

greater degree of significance which we had with the tests run on the basis of absolute values. The changing of the direction of some of the coefficients indicates that some lag in the reaction of LI and L to the independent variables may be present, a force that needs to be examined in future statistical tests.

Crude and tentative as these results are, they seem to indicate that non-member banks behave more procyclically than member banks, an observation that would be corroborated by simple observation of Tables 1,3 and 4, but contradicted by Tables 6 and 7.

b) Non-member and member bank reserve requirements and reserves

The ability of an individual commercial bank to create deposits is limited by (among other things) its reserve requirement. The amount of reserves required to back deposit liabilities varies widely for nonmember banks as compared with member banks. For non-member banks the reserve requirement is set by each state while the Board of Governors sets the requirement for all member banks.

The argument is sometimes made that reserve requirements are typically lower for non-member banks than for member banks and that this is why so many banks prefer to remain outside of the Federal Reserve System. In point of fact reserve requirements for non-member banks are are not typically lower than those for member banks as we can see by an examination of Table 8. As we can see from Table 8, in 1971 there were only fiften states which had reserve requirements on demand deposits for non-member banks which were lower than those set by the Federal Reserve for member banks. Ten states had the same requirements as the Fed, while 25 states actually had higher reserve requirements.

Year

-30

Table 8: Comparison of Reserve Requirements on Demand and

Time Deposits set by the 50 States for Non-member
Banks and those set by the Federal Reserve.*
Selected years, 1962-71

-31

The same pattern held for the reserve requirements on time depoits.

Only two states had reserve requirements on time deposits for non-member banks which were lower than those set by the Federal Reserve for member banks. Sixteen states had the same requirements as the Fed, while 32 states had higher reserve requirements.

Additionally an effort was made to ascertain whether or not the relative level of reserve requirements would be related to the numbers of member versus non-member banks in the various states. No discernable pattern was observed. Of the states where member bank reserve requirements on demand deposits were lower for member banks, in only two, Ohio and Wyoming, did member banks clearly outnumber non-member banks. On the other hand, where reserve requirements on demand deposits were lower for non-member banks, in half the states, member banks were a majority of the number of banks in the state. Thus it appears that if there is a pattern to membership in the Federal Reserve System it is not related to the relative reserve requirements of member versus non

member banks.

Although member banks appear to have an advantage over non-member banks with respect to the absolute level of reserve requirements on deposits, non-member banks can hold these reserves in a wider variety 11/

of assets than member banks.

Whereas member banks must maintain

their reserves in the form of vault cash or reserve deposits at their

district Federal Reserve banks, non-member banks, depending on state laws,

11/

The following discussion is based in part on: Joseph M. Prinzinger and John J. Klein, An Analysis of State Banking Laws and their Relation to Bank Profits. Manuscript in preparation, 1975.