Vol. 8 No. 2 - Article I - Released Date June 1995 - Foot Notes
A 'Two-Window' System for Banking Reform
by Frederick S. Carns
Carns, Frederick S.
Chief, Financial Markets Section, Division of Research and Statistics (DRS). The motivation for this paper arose from work on the FDIC's Report to the Congress on the Findings and Recommendations Concerning the Two-Window Deposit System Proposal, pursuant to Section 321 of Public Law 102-242 (September 1992). Other contributors to the Report were Jay Golter (Division of Research and Statistics), Kathleen M. James (Office of Legislative Affairs), Louise Kotoshirodo (Office of Consumer Affairs), and Lisa M. Stanley (Legal Division). The author thanks Arthur J. Murton for his many helpful suggestions throughout this project. Thanks also to Matthew Billett, Christine Blair, Richard A. Brown, George French, James Freund, and other DRS seminar participants for comments on an earlier draft.
Remarks offered at the Federal Reserve Bank of Chicago's 1994 Conference on Bank Structure and Competition. Parenthetical expressions inserted by the author.
The original concept of two windows was developed by former FDIC Chairman L. William Seidman (1991). However, certain features of the proposal considered in this paper are not attributable to Mr. Seidman. He has not reviewed this version of the two-window approach and should not be held accountable for it. In particular, the heavy emphasis on liquidity transformation in this version reflects the views of the author.
See The Role of Banking in the Financial System and the Economy, Current Issues '94 (FDIC: Washington, DC, 1994) pp. 1.1-1 to 1.1-9. Also, for an evaluation of this premise see French (1994) and Kaufman and Mote (1994).
For purposes of this paper, safety net refers to the triad of federal deposit insurance, discount-window borrowing privileges and federal supervision to maintain bank safety and soundness. In particular, because banks have access to federally insured funding, any activities performed within the insured institution are said to receive safety-net protection.
It will be argued below that this type of intermediation alone merits deposit insurance protection: It is essential for the efficient allocation of credit and yet it is highly sensitive to any threat of bank runs. See Carns, Murton and Nejezchleb (1989) for a more detailed treatment of this topic.
Consolidated supervision refers to the fact that separate units within bank holding companies are supervised as one entity. For purposes of this paper, the essential features of consolidated supervision as currently practiced in the U.S. are two: In general, nonbank firms cannot purchase banks without subjecting their entire organizations to formal federal supervision; and the supervisor sets limits, up to and including prohibition, on activities that may be conducted in nonbank units of the holding company, as well as the insured bank.
A description and evaluation of a more generic two-window proposal can be found in the FDIC's 1992 Report to the Congress on this topic.
Litan (1987), Bryan (1991), and Pierce (1991) are examples of narrow-bank proposals.
There are some important differences among narrow-bank proposals that are neglected for present purposes. In particular, Litan's 1987 proposal does not require the creation of a narrow bank unless the banking organization elects to pursue nontraditional activities. Thus, most generalizations do not apply to Litan's proposal.
See FDIC (1989) and Murton (1989). The fact that there is a defensible rationale does not imply that the benefits of a deposit insurance system will necessarily outweigh the costs in practice.
To clarify, the point is that the characteristics of illiquid loans are idiosyncratic: The credit evaluation process requires costly, specialized information that varies widely across borrowers and their projects. Such information-intensive assets can be distinguished from most residential mortgages, consumer credits and other assets for which the relevant characteristics can be cheaply and accurately conveyed to potential lenders. Information-intensive assets may require special institutional arrangements to ensure adequate funding in the aggregate (such as liquidity-transforming institutions that are protected against runs on their liabilities); no special arrangements are necessary for most marketable assets.
Some liquidity is necessary, but the public purpose of deposit insurance is not primarily to enable institutions to exploit the spreads between marketable instruments and insured deposits, according to this view.
Less-restrictive and operationally less-complex rules for insured bank activities also are compatible with a two-window approach. These are discussed below.
Examples are Tobin (1987), Pierce (1991) and Litan (1987). Unless otherwise noted, affiliate hereafter will refer to any firm within the same corporate structure as the bank, including direct bank subsidiaries, holding-company affiliates, and the like. Also, bank refers to any federally insured financial institution within a holding company or other corporate structure.
This differs from consolidated supervision of the holding company in that the supervisor does not prohibit or otherwise set limits on the nonbank activities of holding-company affiliates. Supervision is along functional lines, so that bank supervisors are concerned about nonbank affiliates only to the extent of their dealings with insured banks.
See Mandate for Change: Restructuring the Banking Industry (Federal Deposit Insurance Corporation: Washington, DC), 1987, Chapter 8. Note: Many elements of the two-window structure are described in Mandate. This article relies heavily on the arguments in Mandate to describe the advantages of a structure that relies upon firewalls and functional supervision.
See Litan (1987), Bryan (1991), and Corrigan (1987).
Despite the declining share of bank loans in short-term business finance, Becketti and Morris (1994) fail to find any evidence that bank loans have become a less important source of business finance on the margin. Stated differently, cyclical reductions in bank loans still constrain business activity despite the availability of nonbank sources of finance such as commercial paper and finance-company loans (p. 1).
See Gorton and Pennacchi (1992) for a counterargument.
A complementary view is that the monitoring of transaction-account activity provides information that is necessary for certain types of commercial lending. Thus, some viable projects may be denied funding in the absence of traditional intermediation by banks. See James (1987), Lummer and McConnell (1989), Billett, et. al. (1993) and Shockley and Thakor (1993).
See Bernanke (1983), Bernanke and Gertler (1987 and 1989), Diamond and Dybvig (1983 and 1986) and the literature cited therein.
See U.S. Department of the Treasury (1991), Chapter 3, Murton (1989), Gorton and Pennacchi (1992), Goodhart (1989), and Becketti and Morris (1992 and 1994).
See FDIC (1987), pp. 60-63 and the references there cited for a more detailed discussion of this issue.
A complication under the two-window system is that the choice of location for fundraising (at the insured-bank, nonbank-affiliate, or holding-company level, or some combination), may be constrained; therefore, diversification on the liability side of the organization may be compromised. However, given that flows of funds from the bank to affiliates are not prohibited, but merely are required to be arm's length, and given that the larger organization is otherwise free to deploy funds as desired, the impact of two-window reforms in this respect should be limited. The author thanks Matthew Billett for raising this issue.
As noted earlier, bank holding companies currently are subjected to limits on the size of any investment in nonbanking businesses. Under the two-window system, owners of banking organizations would be free to shrink the insured bank (consistent with safety and soundness) and reinvest the proceeds in nonbank affiliates as desired.
As with Sections 23A and 23B of the Bank Holding Company Act, firewalls in the two-window system would not prevent the larger organization from supporting the banking unit. The firewalls would be designed to prevent the reverse process whereby the insured institution is weakened for the benefit of its nonbank affiliates.
There is also a legal risk that the bank may be held responsible for the liabilities of its affiliates. For purposes of this article, it is assumed that the law can be changed as necessary to ensure legal separation under a two-window structure.
See FDIC (1992) Appendix for a detailed description of these firewalls and safeguards.
A balanced review and extension of the evidence relating to affiliate relationships are found in French and Hirschhorn (1988). As usual, the evidence is mixed.
See Flannery (1986), p. 224, and Eisenbeis (1983).
See Aharoney and Swary (1983) and Swary (1985). Recall that under a two-window system, some activities currently conducted in the bank may be relegated to nonbank affiliates.
See Flannery (1986), pp. 221-25 for details. The remainder of this paragraph borrows heavily from p. 224 of Flannery (1986) and note 16, and the argument presumes the existence of significant bankruptcy costs as well as impediments to perfect diversification.