The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
February 24, 1984 NOTES ON INSURANCE OF BANK LENDING TO DEVELOPING COUNTRIES by Henry C. Wallich The possibility of an insurance scheme for bank loans to developing countries has been broached from time to time, most recently by Dr. Johannes Witteveen, Chairman of the Group of Thirty, in his 1983 I ! Per Jacobsson lecture. The present note lists a number of considerations, by no means complete, bearing on such an insurance scheme. to explore some aspects of an uncharted area. It was written It should not be interpreted as prejudging the feasibility of such a plan under prevailing circumstances. .1. Prototypes A wide range of credit insurance plans is in operation today. In the United States, mortgages are insured/guaranteed by the Federal Housing Administration and the Veterans Administration, while investments in mortgage pools are guaranteed by the Government National Mortgage Association ("Ginny Mae"), the Federal Home Loan Mortgage Corporation ("Freddie Mac"), and other institutions. Private mortgage insurance, in diverse technical forms, is offered by insurance companies, such as Mortgage Guarantee Insurance Corporation (MGIC) , various banks operating mortgage pools, and others. The low credit risks on mortgages make mortgage insurance an actuarially viable operation. Prepared for the Group of 30. -2- Municipal bond insurance is offered by American Municipal Bond Assurance Corporation (AMBAC) and Municipal Bond Insurance Association (MBIA) , a consortium of five large insurance companies, and municipal bond portfolio insurance by MGIC Indemnity. Some export credit is insured by Export-Import Bank, a government institution, and the Private Export Financing Corporation (PEFCO), a private organization. Some foreign direct investment can be insured, mainly against political risk, by Overseas Private Investment Corporation (OPIC), a wholly owned U.S. government corporation. Bank credit, particularly to individuals, is insured through numerous credit, life and casualty insurers. Commercial and savings bank deposits are insured by the Federal Deposit Insurance Corporation (FDIC), savings and loan association deposits by the Federal Savings and Loan Insurance Corporation (FSLIC), credit-union shares by the National Credit Union Administration, and brokerage accounts through the Securities Investor Protective Corporation (SIPC). Credit insurance, evidently, is widely practiced in the United States. 2. Alternative Objectives of Insurance Insurance can be applied to LDC lending for the purpose both of encouraging such lending and of increasing the safety of banks. It is widely believed that continued lending, although on a more moderate scale, is necessary for the good functioning of the world economy and the health of the international monetary system. It is also believed widely that many banks, especially smaller ones, feel considerable reluctance in that regard. Insurance may ease this problem. Several approaches appear possible. Particular loans could be insured, liabilities of banks could be insured, loan portfolios of banks could be insured. For several reasons, insurance of loans and of deposits seems less feasible than insurance of portfolios. -3- Irisurance of loans tends to create moral hazard. Lenders and borrowers might contract loans, and borrowers might subsequently engage in policies that would inappropriately increase risk. This could perhaps be dealt with by co-insurance, deductibles, partial insurance, and the like. But these devices have costs in terms of the economic effectiveness of the insurance in promoting sound lending. Loan insurance also raises the question of resource allocation. The insuring agency influences the allocation of credit to particular borrowers. Differential insurance premia might be employed to deal with differential risks. They would be difficult to apply, however, for obvious political reasons, if the insuring agency is a governmental organization. Insurance of loans would also discriminate in an undesirable way against borrowers whose obligations can stand on their own feet. The spreads which today measure or should measure the degree of risk of different loans and borrowers work to allocate credit to stronger borrowers. If the insurance scheme were somehow made comprehensive, covering all loans made by a given group of banks, or obtained by a given group of countries, the stronger borrowers would be subsidizing the weaker unless premia are adequately differentiated. The alternative of insuring the deposits of a bank, which is one of the principal insurance features in American banking, seems to lack applicability with respect to bank lending to developing countries. Given the wholesale character of international banks, limitation of deposit insurance to some particular amount — would not be meaningful. in the United States $100,000 The volume of deposits to be insured would, — -4- therefore, be very large. More importantly, while deposit insurance protects the creditor and, by forestalling runs, also the bank, it would not meet the needs of borrowers since banks' willingness to lend to a particular class of borrowers should not be influenced significantly by greater stability of deposits. At the same time, stability of deposits would remove some of the discipline of the market that is needed to limit risk taking. Some "insurance" already is provided to banks through the presence in the market of central banks as lenders of last resort. This protects bank and depositor, not against loss, but against illiquidity. The lender- of-last-resort function is well established, even though in its international aspects, for good reason, not very fully spelled out. Its presence no doubt influences the willingness of banks to lend to developing countries in some nonmeasurable degree, but does not fall under the general heading of credit insurance. The most viable technique seems to be the insurance of portfolios. This would mean to insure a portfolio of LDC loans up to some fraction against loss. To encourage bank prudence, only some large fraction of each loan within the insured part of the portfolio might be insurable, i.e., there would remain an element of co-insurance. No distinction, therefore, would be made between "insured" and "uninsured" loans. Any loan would, in effect, be insured if it should get into difficulties provided that, together with other loans already in troubled condition, it does not cause the share of such loans in a bank's portfolio to exceed the insured fraction. The problems inherent in insuring -5- particular loans, as noted above, would thus be avoided. The main difficulty is to determine the fraction of the portfolio to be insured, so that it would give banks adequate protection without being so large as to be actuarially unmanageable. Banks engage in this measurement of the risk factor in many phases of their operations. The spread over LIBOR is intended to account for risk, in addition to covering the costs of operation and of the capital needed to support the loan. Annual allocations to loan-loss reserves and the level of these reserves are measures of risk. So are the actual loan-loss experience and nonperformance experience. Supervisory classification of loans is an official indicator of risk, but not easily quantifiable. Most of the available indicators of risk suggest that banking risk in general is modest. of their portfolio. Bankers do not expect to lose a significant portion Spreads have frequently been below one percent, and only in some cases, and more recently in reschedulings, above two percent. costs of operation and of capital, only a fraction -- if any — spreads is available to cover risk. Given of these Nevertheless, from an annual charge, a reserve pool can be built up unless the bank chooses to distribute it as dividends. In any event, bankers' policies with regard to spreads on inter- national loans do not suggest any expectation of large loan losses. The loan-loss reserve of large American banks averages somewhat above one percent (1.2 percent) of loans. Their loss ratio on domestic loans reached recent peak levels of .58 percent in 1975 and .56 percent in 1982. The loss ratio on international loans was .30 percent, significantly below the domestic ratio, but with little or no allowance for possible -6- ultimate default on sovereign loans. Nonperforming loans for major banks, as of the third quarter of 1983, were 3.7 percent of loans. Historical experience has, of course, been more favorable. All these indicators, taken by themselves, would seem to suggest that the insured fraction of the portfolio need not be high. Insuring a fraction of two percent might provide a considerable safety factor for a bank, even though, of course, not total protection. These numbers neverthe- less are not a good measure of the degree of uncertainty that many banks are likely to feel with respect to their LDC portfolio. Recent reschedulings often have reflected some kind of irregularity of debt service, even though not rising in all cases to the level of nonperformance, nonaccrual, or regulatory classification as "substandard." The proportion of loans affected by such irregularities may be of the order of 60 percent of LDC loans of American banks. Given that, in the most likely course of events and barring unforeseeable catastrophies, the ultimate loss would be small, nothing like this fraction would require insurance. What is required is that the annual receipts of the insurance fund should, on average, exceed the annual losses, and that the insurance fund should accumulate enough, after paying off losses, to take care of peak-loss years. The insurance may be designed to cover principal only, or interest only, or both. If principal only is covered, which may be subject to long grace periods, substantial funds could accumulate in the meantime from premia, even if eventual benefits are paid promptly instead of after some delay associated with the need to demonstrate that the debtor in fact is defaulting. -7- The amounts to be covered, however, would be large. Hence the fraction of the portfolio that could be insured would be quite small. Insurance may also be designed to cover income, exclusively or in combination with principal. If exclusively, the amounts to be covered would be relatively moderate (compared to principal) and the fraction of the portfolio that could be insured would be larger. Banks1 immediate concern often is a loss of income. For a bank receiving an average interest rate of 12 percent, and a net return of 0.75 percent on assets (all before taxes, taxable- equivalent basis), a loss of interest on a little over 6 percent of its assets would wipe out profits entirely. Many banks, therefore, may prefer to guard against this more immediate risk and face up to risk of loss of principal only over a longer period which in any event can be stretched out by rescheduling so long as interest is paid through insurance or perhaps new loans. If interest is the primary focus of insurance, benefit payments would have to be prompt. to cover principal. Some allowance would nevertheless have to be made The amounts then could become very large. Perhaps insurance of principal could be made contingent on availability of funds after interest (up to the insured fraction of the portfolio) had been paid from the insurance fund. Portfolio insurance, moreover, need not take the form of "first-loss" insurance, several variants of which have been discussed so far. Instead, each of these versions could be cast in the form of catastrophe insurance, such as is familiar in the health-insurance field. Initial losses, of principal or income, could be allowed to be absorbed by the insured. Beyond -8- that, the insurance would take over, in full or in part, up to some limit consistent with the resources of the insurance fund. Insurance of the entire remainder of the portfolio is conceivable, and the risk of such catastrophic losses probably is slight, but nevertheless would be so far beyond the resources of the insurance fund as to lack credibility. It would be very difficult to quantify the proportions of the portfolio representing the lower and upper bounds of the catastrophe-insured part, and no attempt is made here to do so. The possibility nevertheless seems worth exploring if banks should have a preference for catastrophe insurance, as many individuals have with respect to health insurance. Separation of the insured risks into earlier and later maturities would be still another device. This separation frequently is made when commercial banks co-finance with the World Bank, the World Bank then frequently taking the longer maturities. Since risk on the shorter maturities generally is regarded as being lower than on the longer maturities, there could be an opportunity for this approach by an insurance fund if this were the preferred technique on the part of the banks. Further discussion will proceed, however, on the premise that the insurance takes the form of insurance against the first tranche of losses, with some co-insurance. 3. Sources of Funds An effective insurance fund will need a flow of receipts sufficient on average to take care of losses, and beyond that an amount sufficient to generate confidence that even occasional very large losses can be met. The -9- international lending field differs from many others in that actuarial calculations of both magnitudes are very uncertain. It may be possible to create an effective insurance based only on contributions by the insured, i.e., by banks and/or borrowing countries. This is what, in effect, banks have been doing, or should have been doing, in their own strategies of charging for risk and accumulating reserves. amounts to self-insurance. spreads the risk. That activity, however, Pooled insurance is more efficient, because it Accordingly, if the banks have been doing enough in the way of self-insurance, accumulating these risk premia in a pool would provide more than adequate coverage. The possibility must be faced that the banks may not have done enough to guard against a period of high losses. To make up for that, larger contributions or some outside source of funds would be required to establish credible insurance. The additional charges might take the form of a surcharge on the loan, or of some fraction of the spread. The high spreads charged in reschedulings have become controversial, and it might be feasible, as well as appropriate, to channel a large share of them into an insurance pool. The pool would protect only loan portfolios for which a contribution had been made since the inception of the loan. Loans on behalf of which no contribution was made would not be covered. The inclusion or exclusion of rescheduled loans would present a particular problem. To include them would imply a degree of adverse selec- tion, since these loans are by definition weak. On the other hand, most of the potential interest in an insurance plan may well focus on rescheduled -10- loans. The high premia currently being charged on such loans would provide an opportunity to create a rapidly growing insurance fund, if the larger part of them could be channeled into such a fund. Given the presence of insurance, banks might be satisfied with a lower net return. Acceptance of a lower return would also seem advisable in the light of the criticism leveled against reschedulings where fees and spreads have been sharply increased. A weighty objection to the inclusion of rescheduled loans is that it would not serve one of the main purposes of the insurance plan. That purpose would be to encourage new lending (the other being to protect the banks with respect to the consequences of past loans). Limiting the plan to new loans would, of course, make its overall scale much more moderate and probably more manageable. In any event, the possibility that the plan may require outside resources cannot be rejected. If only funds contributed by the banks are available, it could be argued that the insurance would only give the banks their own money back. Of course this would not be so bank by bank. If port- folio compositions differ, some banks would get back more than their own money, others less. That is the nature, and the advantage, of pooled insurance as contrasted with self-insurance. But if the banks in the aggregate have not charged enough of an insurance premium through spreads and fees, an insurance fund derived from contributions based on spreads and fees may not be large enough. An outside source of funds, in any case, whether from government or other sources, should not be a permanent net contribution. it would be a subsidy. If it were, Rather, it should be in the nature of a facility -11- that takes care of peak loads in bad years or that provides an emergency reserve, but it should be paid back over time. Particularly in the early years, before the fund had had time to accumulate adequate reserves, or if losses in the early years should be particularly heavy, such an outside source would be very valuable, perhaps essential. One may recall the "pipeline" into the U.S. Treasury that was established at the inception of the Federal Deposit Insurance Corporation. It was never used, but it gave credibility to U.S. deposit insurance in its early years. Outside funds need not come from governments, but could also be from private sources. If from governments, they need not be funds appropriated specifically for such purpose, but conceivably might be provided by existing official agencies with insurance or guarantee powers, such as the U.S. Export-Import Bank, the U.K. Export Credit Guarantee Department, or the World Bank. While such guarantees would absorb funds that otherwise could be used to finance other LDC credit, use in an insurance plan for commercial bank loans that insured only a fraction of the portfolio would have higher leverage. In order to gain some perspective on the orders of magnitude that might be involved, some purely, illustrative figures are presented. Various observers have estimated that for 1983, $20 billion of net new bank lending (in addition to lending from other sources) would be adequate to cover the borrowing needs of the major borrowing countries. If bank lending proceeded at this rate for five years, with grace periods for each loan of five years, a total volume of new loans of $100 billion would be built up. If one percent premium were paid per year on the outstandings, a total insurance pool of $3 billion would be accumulated (ignoring interest), as detailed in the table below: -12- Premia Receipts (In $ billions) Year Year Year Year Year Maximum Interest Losses Annual Cumulative Annual $0.2 0.4 0.6 0.8 1.0 0.2 0.6 1.2 2.0 3.0 1.5 3.0 4.5 6.0 7.5 one two three four five Cumulative 1.5 4.5 9.0 15.0 22.5 These $3 billion would be available to meet defaults on either principal or interest. With respect to principal defaults (which could occur only after the five-year grace period), the fund is very small, although losses of 3 percent exceed by far any normal loss ratio in domestic banking. With respect to defaults on interest, if $20 billion are loaned per year at 10 percent, and if coinsurance of 25 percent is assumed, maximum annual losses to the fund would be $1.5 billion in the first year, rising to $7.5 billion in the fifth, and cumulative potential losses would be $22.5 billion after five years, ignoring that total nonpayment of interest as here assumed would make continued annual lending at $20 billion per year unlikely. Thus for each year, and for the five-year period as a whole, the insurance fund would have resources equal to 13.3 percent of the maximum interest loss chargeable to the fund. An accumulation of premia of $3 billion over five years is modest, to be sure, whether matched against total LDC lending, or lending to one of the largest debtors, or the potential guarantee capacity of some official institutions. On the other hand, a fund that (after coinsurance of 25 percent) Would be able to insure 13 percent of the new loans portfolio in the case of income loss, or, alternatively, up to 4 percent in the case of principal l°ss, would not seem negligible in relation to historical average loss experience of banks. -13- If outside sources of funds are regarded as necessary, both national and international sources could be considered. National export credit insurance or guarantee organizations have already been noted. The International Monetary Fund is not authorized by its statutes to provide insurance or guarantees, but might be able to establish a trust fund, such as was done in other contexts, that could do so. Contributions to such a fund presumably would have to come from outside the International Monetary Fund's own resources. The World Bank is authorized to give guarantees. These, as already noted, absorb lending capacity dollar for dollar, but the leverage with respect to the volume of bank lending that could thus be induced might be much higher. Any contributions from national or international sources should not be designed as subsidies, but as temporary resources that lend credibility to the insurance plan and that should be repaid by the insurance plan if drawn upon. Private insurance companies, finally, could also be thought of as outside sources. Particularly if reinsurance were involved, this would not be altogether out of line with their business. But they could not rely on actuarial experience derived from that business, as far as LDC risk was concerned. Indeed, if such an insurance scheme were viable, consideration of the potential role of private insurers raises the question why the market has not already produced such a scheme. However, while high risks may be an adequate explanation, the lack of demand for such a scheme until quite recently, and the possibly negative attitude toward it on the part of many banks even now, may also have been a reason. -14- Finally, consideration might be given to national rather than international insurance schemes. These would be easier to negotiate. However, if they required substantial contributions from the banks of the respective country or from their customers, those banks might find themselves at a competitive disadvantage. 4. Sponsorship and Management Sponsorship of the insurance plan, and its continuing management, presumably would depend on the source of funds. If the sole or principal source were to be contributions by the banks, in proportion to covered loans, a banking syndicate or some bank-related group would be the logical sponsor. Syndications and reschedulings have created working relationships among large international banks that may make such an arrangement possible. If substantial contributions from governments were involved, the organization presumably would have to have official character. Possibly, the IMF and the IBRD, whether they supply the resources or not, could serve as initiators and administrators. If existing insurance organizations were to participate on a large scale, protection of their interests would require participation in management. j,. Other constellations, of course, are conceivable. Participation in the Plan How general should the participation of banks and borrowers be? Is it conceivable that, without legislation or official agreement, all LDC loans could become subject to such a scheme? If universality were aimed at, would it place a burden on strong borrowers and strong banks for the benefit of the rest? m If participation were voluntary, both for banks and borrowers, ight it create opportunities for free-riders, or for the development of -15- uncovered weak spots? What would be the means by which some kind of order and coordination could be achieved? Could governments, even in the absence of firmer arrangements, encourage participation by more favorable treatment of insured loans in bank supervision? Could appropriate arrangements be made with respect to country classification, loan classification, capitaladequacy requirements, provisioning requirements? Could the IMF and the IBRD exert an influence toward participation? What kinds of loans would be subject to such a scheme? LDC loans, East-Bloc loans, small developed-country loans, all international loans? Should there be differential treatment for syndicated loans, trade credit, guaranteed credit, shorter and longer portions of syndicated loans, interbank loans, offshore-center placements, and for rescheduled loans, if included? Should there be country limits, and how would they be established and administered? Should there be special treatment for countries that in a regulatory framework had received adverse classification? supervisory authorities be involved in some form? Should national None of these matters seems decisive, but they need to be addressed. _6. Nature and Timing of Compensation of Losses As noted before, some co-insurance by insured banks would be desirable to ward off moral hazard. the above illustrative calculations. A portion of 25 percent was assumed in Such a level of coinsurance ought to be sufficient to guard against any temptation either to allow losses to increase up to the insured level, or to weight the scales in favor of accepting default instead of refinancing. -16- Given this, what is the appropriate time for recognition or pay-off on a loss? Insurers in other fields have established routine practices that may provide guidelines. In the case of some forms of insurance, failure to receive payment of interest and principal on a timely basis gives rise to an immediate claim. In others, there may be a protracted process of documenting the nature of the loss, its definitiveness, or possible redress open to the insured. matter is apt to be particularly difficult. For sovereign risks, the Lack of perfect punctuality in the service of principal and often even interest has become a not unusual event. Should the insurer of a sovereign credit subrogate himself immediately for the lender and pay off the latter? Banks in many cases, as already noted, may be more concerned about timely receipt of interest than about ultimate repayment because of the implications of nonaccrual status of loans. Moreover, determination of ultimate loss in the case of a sovereign borrower may be virtually impossible. Even repudiation need not be definitive, since a successor government may resume payments. In the case of sovereign loans, judgmental provisioning, voluntary or mandated by supervisors, may be the most tangible evidence of "loss." Resolution of the definition of "loss," in turn, may bear upon the question who is to act with respect to the defaulting debtor. Representation of continued creditor interests could remain in the hands of the banks involved, or it could pass to the insuring agency. J. Probable Attitudes of Banks For lending banks, an insurance scheme has attractions but also drawbacks. Among the advantages are the greater risk protection at least for those banks that have done no more than average provisioning. Whether -17- the insurance plan is additional to provisioning, or in part a substitute, pooled insurance in any event is more efficient than self-insurance. Insurance might be welcomed by banks as protection against the extreme competition that during the late 1970!s brought spreads down to levels inadequate to cover risk. Channeling part or all of the spread into an insurance fund would allow these resources to accumulate untaxed, whereas in most countries they would be taxed if taken into income. Banks may expect that the ultimate burden of insurance can be shifted to the borrower. They may be attracted by the prospect of more liberal supervisory treatment of loans in an insurance plan. The IMF may take a more positive view of loans that are in an insurance plan. Borrowing countries may be under better discipline if their loans are in an insurance plan. Indeed, some firmer relation to the IMF than is provided by its present practices and statutes, such as commitment in the loan-contract to go to the IMF at an early point of a problem situation, might be made part of a loan and insurance contract. Better regulatory treatment, tax exemption of the insurance premium, and a close IMF relation might be important plusses for banks in addition to the insurance itself. On the other hand, banks may find much to object to in an insurance plan. There is the cost to bank or customer. There is the fact that insurane can never provide complete protection, short of some legislative commitment. Strong banks may believe that they are better off accumulating their own reserves. They may see weaker banks becoming more competitive by virtue of insurance. Problems may be seen in the appearance of free-riders, mavericks, and other nonparticipants. -18- 8. Attitudes of Borrowers While the insurance plan should benefit borrowers by encouraging banks to lend, these borrowers may be concerned about possibly higher costs. A market analysis may lead to the conclusion that costs diminish under pooled insurance, but this would be true only if banks acting independently were in the practice of establishing adequate reserves. concerned about pressure coming from the insurer. Borrowers might be They might be concerned about substitution of the insurer for the bank in case of loss, if the plan so provides instead of leaving administration of the delinquent loan in the hands of the bank. Declaration of default by the insurer might loom as a threat to the borrower. If participation is a matter of choice for the borrower, he faces additional problems. 9. Broader Economic Implications of an Insurance Plan Insurance plans may have secondary repercussions that are not intended but would nevertheless be damaging. discipline of the market. Insurance might reduce the It might lead to subsidization, of the weak by the strong, or of private interests by the public. to "adverse selection." It might be vulnerable It might be viewed as a "bail-out" for banks. It might inappropriately change the allocation of credit from what the market would bring about. On the other hand, these dangers seem avoidable, through the proper structuring of the plan, through charging of differential rates and, in general by taking these concerns into account in the design and the administration of the plan. The basic principle of risk-pooling is in any event well established has widely been recognized as beneficial, and should be applicable also to the field of international bank lending. #