The Model. While the historical record shows that the Federal Reserve attempted to use the nonrate terms of access to control discount window borrowing in the 1920s and early 1930s it is not clear the attempts were successful. Assessing whether changes in credit policy had a material effect on bank borrowing requires empirical tests using a model that can measure the administrative pressure applied at the discount window. A useful starting point in developing such a model is in specifying the costs to banks in meeting a reserve need. Cost-minimizing banks would weigh the cost of borrowing from the discount window against the cost of liquidating assets or borrowing from an alternative source. The cost of borrowing from the Federal Reserve is the interest paid at the discount rate, Rd, plus the implicit costs of supervisory surveillance. These costs are those resulting from the administrative pressure supervisors apply to discourage banks from engaging in arbitrage when discount rates were below market rates of interest, as they were for most of the 1920s and many months in the early 1930s. In this model the implicit costs are represented by a borrowing function in which the marginal surveillance cost of access to the discount window, c(B/K), rises with the amount borrowed, B, relative to bank capital, K. Capital is the appropriate scale variable given that Federal Reserve banks had from the start focused on borrowing relative to capital when seeking to restrain bank borrowing (▇▇▇▇▇▇▇▇▇▇▇, 1932, p. 44). The implicit costs may be thought of as the opportunity costs of providing the collateral for the borrowings or the capital adjustments required by supervisors whose attention is drawn to the bank by the borrowing. The cost of liquidating the assets is the foregone interest on the assets, RA, plus the transactions costs incurred in selling the assets, tc, measured as a portion of the value of the assets. Although a small market for federal funds developed in several cities in the 1920s in which banks could borrow or lend reserves, most banks met reserve drains by liquidating assets — particularly their holdings of call loans and short- term marketable securities (▇▇▇▇▇▇, 1938, 93-97 and ▇▇▇▇▇▇, 1964) 3-13). The total cost, C, to banks of meeting their reserve need, RN, can be expressed as: where (1) C = RdB + c(B/K) + RaA + tcA Rd is the discount rate, c(B/K) is the implicit or surveillance costs involved in borrowing, Ra is the interest rate on the alternative source of funds, For estimation it is necessary to commit to a functional form for the implicit surveillance cost function. One such function in which surveillance costs rise with the level of borrowing relative to capital is: (2) c(B/K) = (λK/2)(B/K) 2 , c’(B/K) = λ(B/K) where λ > 0 measures the degree of administrative or surveillance pressure on the bank. Substituting this explicit cost function into (1), scaling all variables by bank capital and minimizing C subject to the constraint that RN = A + B, letting Spread = Ra - Rd , and solving for B/K, gives the bank’s demand for borrowed reserves as: (3) B/K = (1/λ)(Spread + tc), B/K > 0. The demand for borrowed reserves rises with the spread between markets rates and the discount rate, Spread, rises with the transactions costs incurred in obtaining reserves from an alternative source, tc, and falls with an increase in administrative pressure at the discount window, λ. The model described by equation (3) is a static in that describes only current-period borrowing. The Federal Reserve’s repeated avowal to discourage continuous borrowing during the 1920s suggests the implicit costs should be specified as a function of both current and past borrowing, which would make the demand for borrowed reserves the solution to a dynamic optimization involving current, past, and expected future borrowing as in ▇▇▇▇▇▇▇▇▇▇ (1983). Considerable evidence exists, however, that the sanctions against continuous borrowing were not enforced effectively in the interwar period. A study commissioned by the Federal Reserve found that as of August 1925 found more than 588 member banks had been borrowing steadily for a year or more from the Federal Reserve. Of these, 239 had been borrowing since 1920 and 122 had begun borrowing before that; see ▇▇▇▇▇ (1971) pp. 34-35. One Federal Reserve bank was reported to have permitted its members to renew their 15-day loans indefinitely, and as of 1928 had never refused to renew, with some loans running as long as four years (▇▇▇▇▇▇▇▇▇▇▇, 1932,
Appears in 1 contract
Sources: Research Paper
The Model. While the historical record shows that the Federal Reserve attempted to use the nonrate terms of access to control discount window borrowing in the 1920s and early 1930s it is not clear the attempts were successful. Assessing whether changes in credit policy had a material effect on bank borrowing requires empirical tests using a model that can measure the administrative pressure applied at the discount window. A useful starting point in developing such a model is in specifying the costs to banks in meeting a reserve need. Cost-minimizing banks would weigh the cost of borrowing from the discount window against the cost of liquidating assets or borrowing from an alternative source. The cost of borrowing from the Federal Reserve is the interest paid at the discount rate, Rd, plus the implicit costs of supervisory surveillance. These costs are those resulting from the administrative pressure supervisors apply to discourage banks from engaging in arbitrage when discount rates were below market rates of interest, as they were for most of the 1920s and many months in the early 1930s. In this model the implicit costs are represented by a borrowing function in which the marginal surveillance cost of access to the discount window, c(B/K), rises with the amount borrowed, B, relative to bank capital, K. Capital is the appropriate scale variable given that Federal Reserve banks had from the start focused on borrowing relative to capital when seeking to restrain bank borrowing (▇▇▇▇▇▇▇▇▇▇▇, 1932, p. 44). The implicit costs may be thought of as the opportunity costs of providing the collateral for the borrowings or the capital adjustments required by supervisors whose attention is drawn to the bank by the borrowing. The cost of liquidating the assets is the foregone interest on the assets, RA, plus the transactions costs incurred in selling the assets, tc, measured as a portion of the value of the assets. Although a small market for federal funds developed in several cities in the 1920s in which banks could borrow or lend reserves, most banks met reserve drains by liquidating assets — particularly their holdings of call loans and short- term marketable securities (▇▇▇▇▇▇, 1938, 93-97 and ▇▇▇▇▇▇, 1964) 3-13). The total cost, C, to banks of meeting their reserve need, RN, can be expressed as: where
(1) C = RdB + c(B/K) + RaA + tcA Rd is the discount rate, c(B/K) is the implicit or surveillance costs involved in borrowing, Ra is the interest rate on the alternative source of funds, For estimation it is necessary to commit to a functional form for the implicit surveillance cost function. One such function in which surveillance costs rise with the level of borrowing relative to capital is:
(2) c(B/K) = (λK/2)(B/K) 2 , c’(B/K) = λ(B/K) where λ > 0 measures the degree of administrative or surveillance pressure on the bank. Substituting this explicit cost function into (1), scaling all variables by bank capital and minimizing C subject to the constraint that RN = A + B, letting Spread = Ra - Rd , and solving for B/K, gives the bank’s demand for borrowed reserves as:
(3) B/K = (1/λ)(Spread + tc), B/K > 0. The demand for borrowed reserves rises with the spread between markets rates and the discount rate, Spread, rises with the transactions costs incurred in obtaining reserves from an alternative source, tc, and falls with an increase in administrative pressure at the discount window, λ. The model described by equation (3) is a static in that describes only current-period borrowing. The Federal Reserve’s repeated avowal to discourage continuous borrowing during the 1920s suggests the implicit costs should be specified as a function of both current and past borrowing, which would make the demand for borrowed reserves the solution to a dynamic optimization involving current, past, and expected future borrowing as in ▇▇▇▇▇▇▇▇▇▇ (1983). Considerable evidence exists, however, that the sanctions against continuous borrowing were not enforced effectively in the interwar period. A study commissioned by the Federal Reserve found that as of August 1925 found more than 588 member banks had been borrowing steadily for a year or more from the Federal Reserve. Of these, 239 had been borrowing since 1920 and 122 had begun borrowing before that; see ▇▇▇▇▇ (1971) pp. 34-35. One Federal Reserve bank was reported to have permitted its members to renew their 15-day loans indefinitely, and as of 1928 had never refused to renew, with some loans running as long as four years (▇▇▇▇▇▇▇▇▇▇▇, 1932,
Appears in 1 contract
Sources: Research Paper