Grants versus Loans for Development Banks

In recent years, economists have increasingly debated whether multilateral development banks, such as the World Bank, should switch from making subsidized loans to giving outright grants. It is no small question. The combined loans of the World Bank Group and brethren regional entities such as the Asian and Inter-American Development Banks, approach $300 billion. Their funds constitute a main channel through which rich country governments provide assistance to developing country governments. In Bulow and Rogoff (1990), we first developed the case for a shift to outright grants. We argued that under the status quo, a vastly disproportionate share of aid goes to middle income countries via disguised interest subsidies, rather than to the poorest countries. We also argued that a shift to grants would protect donor banks from sometimes having to play a “bad cop” role when trying to collect net repayments rather than fully rolling over loans. The “Meltzer Commission” (International Financial Institution Advisory Commission, 2000) report on government sponsored international lending institutions famously took a similar view. Supporters of the status quo often argue that development bank loans to middle income countries are in fact highly profitable, and are essential for allowing institutions like the World Bank to subsidize aid to poor countries. We shall argue that the Bank’s profitability is an accounting artifice that greatly underestimates the risks of the Bank’s portfolio. Another argument for loans is that multilateral development banks have a superior enforcement technology that helps international debt markets to function more efficiently. Thus loans allow financially strapped governments, including in middle-income countries, to borrow more than they could otherwise. We will argue that this benefit, too, is an illusion. In those cases when official lending does expand a developing country government’s borrowing capacity, it effectively enables the government to commit the country to repayment levels beyond that supported by domestic political consensus, creating moral hazard for shortsighted rulers. In theory, better credit access to finance, say, public infrastructure projects can be highly beneficial. In practice, however, the increased risk of debt crisis all too often outweighs any gain ordinary citizens might enjoy from the loans. Furthermore, moral hazard on the part of lenders, who may be able to induce rich countries into subsidizing the bailout of troubled middle-income borrowers, may mean that aggregate lending is excessive even if multilaterals merely displace equivalent private debt. We do not argue for eliminating assistance to middle-income countries. On the contrary, we would favor expanding aid in general, albeit in far greater proportion to the world’s poorest countries. Note that in principle, any country with market access could use grant flows to help defray interest rate costs on loans if it so chose, but development banks would never need to assume a “bad cop” role in enforcing debt.


Grants versus Loans for Development Banks
By JEREMY BULOW AND KENNETH ROGOFF* In recent years, economists have increasingly debated whether multilateral development banks, such as the World Bank, should switch from making subsidized loans to giving outright grants. It is no small question. The combined loans of the World Bank Group and brethren regional entities such as the Asian and Inter-American Development Banks, approach $300 billion. 1 Their funds constitute a main channel through which rich country governments provide assistance to developing country governments.
In Bulow and Rogoff (1990), we first developed the case for a shift to outright grants. We argued that under the status quo, a vastly disproportionate share of aid goes to middle income countries via disguised interest subsidies, rather than to the poorest countries. We also argued that a shift to grants would protect donor banks from sometimes having to play a "bad cop" role when trying to collect net repayments rather than fully rolling over loans. The "Meltzer Commission" (International Financial Institution Advisory Commission, 2000) report on government sponsored international lending institutions famously took a similar view.
Supporters of the status quo often argue that development bank loans to middle income countries are in fact highly profitable, and are essential for allowing institutions like the World Bank to subsidize aid to poor countries. We shall argue that the Bank's profitability is an accounting artifice that greatly underestimates the risks of the Bank's portfolio.
Another argument for loans is that multilateral development banks have a superior enforcement technology that helps international debt markets to function more efficiently. Thus loans allow financially strapped governments, including in middle-income countries, to borrow more than they could otherwise. We will argue that this benefit, too, is an illusion. In those cases when official lending does expand a developing country government's borrowing capacity, it effectively enables the government to commit the country to repayment levels beyond that supported by domestic political consensus, creating moral hazard for shortsighted rulers. In theory, better credit access to finance, say, public infrastructure projects can be highly beneficial. In practice, however, the increased risk of debt crisis all too often outweighs any gain ordinary citizens might enjoy from the loans.
Furthermore, moral hazard on the part of lenders, who may be able to induce rich countries into subsidizing the bailout of troubled middle-income borrowers, may mean that aggregate lending is excessive even if multilaterals merely displace equivalent private debt.
We do not argue for eliminating assistance to middle-income countries. On the contrary, we would favor expanding aid in general, albeit in far greater proportion to the world's poorest countries. Note that in principle, any country with market access could use grant flows to help defray interest rate costs on loans if it so chose, but development banks would never need to assume a "bad cop" role in enforcing debt.

I. Why Non-crisis Lending Does Not Add Value
We have already mentioned the two principal rationales for loans: (1) Multilaterals claim to have a superior debt enforcement capacity that enables them to expand a country's debt capacity, * Bulow: Graduate School of Business, Stanford University, Palo Alto, CA 94305 (e-mail: jbulow@ stanford.edu); Rogoff: Department of Economics, Harvard University, 232 Littauer, Cambridge, MA 02138 (e-mail: krogoff@harvard.edu). 1 Using data from their most recent annual reports, total loans outstanding for the largest multilateral development banks include: International Bank for Reconstruction and Development ($105 billion), the International Development Agency ($70 billion-after taking out explicit grant component), International Finance Corporation ($15 billion), Inter-American Development Bank ($50 billion), the Asian Development Bank ($25 billion), the European Bank for Reconstruction and Development ($8 billion), and the African Development Bank ($8 billion). We exclude the $90 billion loaned by the International Monetary Fund in this discussion, as the IMF's stipulated role is to promote global financial stability rather than to provide long-term development assistance, and it is not designed to be a development bank. Note that most multilateral development bank loans are government to government loans though a small percentage go to the private sector.
and (2) The multilaterals' middle-income country loan business has been a long-run money maker, generating profits that can be used to help developing countries. A third rationale is that by making large scale loans, development banks are able to obtain far greater leverage over developing country policies than they would be able to with a steady small stream of grants. Thus, loans are a way of leveraging the donors' influence over the country.
Setting aside the valuable technical assistance that multilateral lenders can provide (and which we would continue to finance with grants), a great body of evidence suggests that one should be extremely cautious in using an external enforcement mechanism to expand the borrowing capacity of a developing country whose own domestic institutions and governance are too weak to support a high volume of international debt. Countries with weak institutions are prone to serial default, often resulting in economic chaos and sustained pauses in growth (see Carmen Reinhart and Rogoff, 2004). Indeed, in Bulow and Rogoff (1990), we argued more broadly that rich country governments should also be cautious in allowing their legal systems to be used to enforce sovereign debts of countries whose own legal institutions are too malleable or corrupt to offer foreign lenders meaningful recourse in the event of default. Even where corruption per se is not an issue, leaders may face short-term pressures to borrow maximally, ignoring heightened long-term risks of a debt crisis. Just as aid is arguably more effective in countries with good governance and strong institutions (e.g., Craig Burnside and David Dollar, 2000), the same is likely true for expanded debt capacity.
What about the argument that profits on multilaterals' middle-income country loan business helps finance all the useful technical assistance they provide? Whereas we do not dispute the need for high quality technical assistance, one has to question this financing mechanism if pre-crisis multilateral loans systematically exacerbate the depth and severity of middleincome country financial crises. The tragic Argentine debt crisis of 2001 provides an illustration of this problem; the World Bank and Inter-American Development Bank lent billions when Argentina's government was overborrowing during the 1990s.
In any event, it is highly debatable whether multilateral lending banks actually do earn an economic profit. Bank accounting attributes little risk to "hard money" development loans. Some may argue that the multilaterals do face only minimal risk on middle income country loans, since rolling over troubled loans has thus far avoided the need for large write-offs. But recall that the U.S. federal deposit insurance system also made a "profit" on its premiums for over fifty years before losing well over $100 billion. There is no reason to believe that multilateral development banks will perpetually avoid a similar fate.
A closer look at the balance sheet of the World Bank's main lending arm, the IBRD (International Bank for Reconstruction and Development) helps illustrate the problem. The IBRD reports a 29 percent equity to loans ratio and a 3 percent bad debt reserve, which would appear to put it in rock solid financial condition. These ratios, however, value Bank loans at book value, rather than at market value using private debt prices. In mid-2003, the Congressional Budget Office estimated that multilateral debt was actually worth only 50 cents on the dollar, under the assumption that, de facto if not de jure, multilateral debt is only of equal priority with loans from private lenders (see Douglas Holz-Eakin, 2004). The CBO calculation admittedly runs against the official line that multilateral debt is senior, justifying the multilaterals' practice of charging interest rates far below those charged by private lenders. However, the CBO's calculations are quite consistent with empirical tests of seniority presented in Bulow et al. (1992), who find they cannot reject the hypothesis that multilateral debt is of equal priority to private debt. Under the assumption of equal priority, the World Bank would have had negative net worth at the end of 2004, even after the sharp narrowing of risk spreads on developing country debt that took place during that year.
If the World Bank really has negative net worth, what then can explain the AAA market rating the Bank gets on the bonds that it issues to help finance its lending and technical assistance activities? The answer is that the Bank has a call on the capital of its rich shareholder countries that provides private bondholders with the equivalent of deposit insurance. Indeed, the Bank carefully constrains its develop-ing country loans to about half its statutory limit, ensuring that its bond issuances never exceed the sum of rich country callable capital plus its liquid financial assets (held mainly in rich country securities). As Table 1 illustrates, even if the Bank's developing country loan portfolio became worthless overnight, it would still have $25 billion more in liquid assets and guarantees than in debts.
Even if one ignores the obvious risks posed by future defaults, the Bank's profits reported to date may still be overstated, as indeed may be its status as a creditor senior to private lenders. In practice, countries that balk at repaying large net debts to the World Bank are often coaxed into doing so with the explicit or implicit promise of higher future bilateral loans, aid, or both. Nancy Birdsall et al. (2003) and Bulow et al. (1992) both find evidence in favor of such webs of sidepayments. A clear example is the current use of grants and subsidized loans from other official lenders to prevent default on the hard money loans of the African Development Bank.
Finally, what of the argument that jumbo size development loans give donors superior leverage to a subsidy-equivalent level of grants? No doubt big loans do give donors more up front leverage. Once the loans are committed, however, the Bank is often left with very little leverage in future periods when funds are supposed to be reflowing from the borrower. The issues for other development banks are similar.

II. The Benefits of Moving to Grants
We have already highlighted how grants give much greater transparency by unbundling market rate loans, grants or subsidies, and technical assistance. Empirical estimates suggest that the subsidies inherent in middle-income country loans are potentially quite large (see Bulow et al., 1992). With grants, any implicit loan subsidies would be made explicit and, in principle, aid could be channeled more fairly and efficiently.
Another reason for moving to grants is to minimize "loan pushing." Multilateral development banks sometimes have their own internal pressures to pump out loans, inducing politically fragile developing countries to take on unwanted debt. This can happen for example, when a multilateral offers to make a loan to a state government, which will then politically pressure the federal government into ratifying and guaranteeing the multilateral's "cheap" debt.
Grants eliminate the needless tying of a project's financing to the amount of subsidy provided. At the end of 2004, China possessed over $500 billion in reserves. If the World Bank wishes to support new schools or hospitals why should this only be possible if the Chinese agree to borrow money?
The replacement of official loans with grants will not eliminate the pressure to bail out private creditors during debt crises, but at least the nature of any such transaction will become more transparent.

III. Transition and Financing
Given the status quo, how can one phase the World Bank, or any of the regional development banks, out of the lending business? There are a number of possible schemes. We offer below a highly plausible three-step plan for the World Bank's IBRD arm, which lends mainly to middle-income countries. (The plan can easily be generalized to incorporate the Banks' other lending arms as well.) Our approach would both transition the IBRD out of loans and provide a firm foundation for future grant disbursements: (1) No new loans to developing countries.-At present, ninety percent of the face value of IBRD debt is scheduled to come due within ten years. The Bank could reschedule or write down loans as needed, but any cash transfers to a country would be in the form of grants. (2)   The purpose of (1) is to ensure a smooth transition to a grant system, while not forcing the Bank to sell any of its interest in debt claims it may regard as senior but that the market may regard as less valuable.
While (1) would reduce the Bank's undiversified exposure to middle-income developing country credit risk, (2) would further reduce risk by de-leveraging the "World Bank Hedge Fund" (see Table 2).
The goal of proposal (3) is to strengthen the Bank's endowment for providing future grants and technical assistance. Indeed, when combined with the elimination of further borrowing, the new financial structure of the World Bank would ultimately be similar to that of a private $100 billion foundation, the exact value depending on its success in collecting its preexisting loans. Note that the above plan would share the burden of financing the new grant agency between developing countries, some of whom would be making loan repayments, and rich countries, who would be donating additional endowment funds.

IV. Conclusions
The long-standing practice of bundling loans into multilateral development assistance is an anachronism that no longer makes sense in to-day's highly liquid capital markets, particularly in the case of middle-income (emerging market) countries. In cases where multilateral loans significantly expand chronic defaulting countries' borrowing capacities, they may well be counterproductive. Stronger international enforcement of developing country loans largely encourages greater borrowing by exactly those kinds of weakly governed states for whom expanded borrowing is least likely to be productive, and most likely to lead to debt crises. Even if official lending mainly crowds out private lending, bundling aid in the form of subsidized loans makes aid flows highly non-transparent. This opacity potentially distorts the pattern of aid flows across countries, with middle-income countries receiving a much larger share of total multilateral development aid than is commonly realized. We reject various standard arguments for preferring loans to grants, such as needing to have big programs to induce borrower compliance. We offer a relatively simple and practical mechanism for achieving a transition from loans to grants. Our proposal is designed to be market friendly and to put funding for multilaterals' future grant and technical assistance activities on a firm and predictable basis.
We have focused on multilateral development bank loans but most of our analysis applies with even greater force to the broad universe of bilateral official lenders, who collectively have more than $450 billion in loans outstanding, 50 percent more than the multilateral development banks. Bilateral loans share all the problems of multilateral loans, plus they typically attach strings that siphon off resources to favored rich country industries. Nevertheless, the case for shifting to a grants-only multilateral regime seems compelling regardless of how quickly the individual countries guide their bilateral lending arms to move in the same direction.