Wednesday, 12 March 2014

When Do We Have a Donor Dilemma?

Considering the wide range of scenarios for future developing-country growth and poverty reduction, prudence would suggest to approach strategic changes of multilateral concessional windows with a precautionary, rather than a deterministic, perspective to enable flexible institutional response. That perspective should not only watch the prospect of future graduation for countries today eligible for multilateral soft windows, but also the prospect of reverse graduation resulting from disasters, military conflict and governance failures[1]. Don´t allow the shareholders of multilateral soft windows “to sleepwalk into the future” (Séverino & Moss, 2012)[2] - yes; but don´t either allow them to ignore the option value of preserving the financial and institutional strength of the soft windows by “declaring success” and letting them shrink.  After all, the configuration of poverty scenarios and strategic options is such that there may be a multilateral donor dilemma; but then there may be not...  Consider Table 1 and try to attach probabilities to the sum of outcomes that would constitute a donor dilemma (left column) and to those where a donor dilemma would not exist (right column).


Table 1: When Do We Have a Donor Dilemma?

Poverty Gap/Donor Dilemma
Poverty Gap/No Donor Dilemma
Closed in LICs & MICs
In ´stable´ LICs
Only in ´fragile´ LICs
In MICs with little
capacity for redistribution
Also in MICs with good
capacity for redistribution
Redefinition of eligibility status
and of IFI mandates


Case 1: There will be a multilateral donor dilemma when “we beat” poverty by 2025 without changing IFI mandates. History suggests that the absence of institutional change is unlikely for a number of reasons: first, the concept of ODA-eligibility is ´elusive´ (Schlögl, 2013)[3] as it does not have hard legal force; second, there have been exceptions (such as currently the small islands, or Israel before); third, rent-seeking by affected bureaucracies; and fourth,  divergent paradigms with respect to poverty (absolute vs relative).

Case 2: This is the old, Paul Collier´s (2007) Bottom Billion. Should poverty be concentrated in a few countries labelled ´fragile´ by 2025[4], aid volume targeted at poverty would have to shrink as the absorptive capacity in those countries is too limited relative to current aid volumes. Except for small island economies, there are 25 countries currently in that category, of which 20 in Africa (OECD, 2014, The Where of…, op.cit, Figure 2), with only 24% of the world poor living there. Moreover, one third of this country group is considered resource rich, intensifying the donor dilemma.

Case 3: It cannot be excluded that the majority of the world´s poor will live in MICs by 2025. This is the new, Andy Sumner´s (2013)[5] Bottom Billion. They constitutes another side of the donor dilemma, mainly because of the widespread presumption that MICs can take care of themselves, given their tax and redistribution capacity, but also their other assets (FX reserves, sovereign wealth funds) and their own donor activities.

Case 4: This is the ´easy´ case for donors, as the category of ´stable´ LICs is thought to have the governance and institutions that allow for efficient aid allocation and absorption. The problem is that there are not many currently in that category, according to the OECD: 15. But it could be that aid to the fragile LICs catalyzes governance reform so that some of these countries move to ´stable´ before graduation.

Case 5: This is the ´uneasy´ case for donors as it has become increasingly difficult to explain to voters at home. Still, a large part of chronic poverty is likely to be located in MICs that have only little capacity to redistribute. These are the MICs with annual consumption per capita under $2,000; the tax burdens required to close the national poverty gaps are prohibitive in these countries, according to Ravaillon (2012). Kanbur and Sumner (2011)[6] provide some arguments for aid to poor people in MICs rather than only to poor countries: chronic poverty calls for common humanity[7]; concessional finance facilitates working on GPGs with strong externalities for donors; and creating a common knowledge base for policy lessons to combat poverty in poor countries. While uneasy for donors, these MICs do not necessarily constitute a donor dilemma.

Case 6: Strategic options exist for the shareholders of IFIs to attenuate the donor dilemma (if there is one) up to 2025. These options will be shortly presented and discussed: complementing/redefining the current IDA cutoff (GNI/capita); introducing subsovereign allocation; smoothening transition periods from IDA-only via blend status to IBRD-only (and correspondingly for the other MDBs); and opening the soft windows for global public goods.

[1] For example, since IDA´s inception, 36 countries have graduated, of which 11 became ´reverse graduates´ subsequently; another 17 IDA-only countries were, at one point in time, assessed as creditworthy for IBRD financing and classified as a blend country, but subsequently reversed to IDA-only status. For detail, see IDA (2012), “Review of IDA´s Graduation Policy”, World Bank: Washington, DC, October.
[2] Jean-Michel Séverino and Todd Moss (2012), Soft Lending Without Poor Countries, Center for Global Development: Washington, DC, October.
[4] As defined by OECD (2014), The Where of Development Finance, OECD: Paris, January.
[5] Andy Sumner (2013), „Where Do The Poor Live?”, World Development, Vol. 40(5), pp. 865-877.
[6] Kanbur, Ravi and Andy Sumner (2011), “Poor Countries or Poor People? Development Assistance and the New Geography of Global Poverty”, CEPR Discussion Paper 8489, CEPR: London.
[7] The German philosopher Thomas Pogge (Yale) is a prominent representative of that school of thought which has challenged Rawls´ A Theory of Justice that applies only within a nation state: See Ravi Kanbur (2014), Resetting IDA’s Graduation Policy, Cornell University, mimeo, January.

Monday, 10 March 2014

The Concessional IMF Facilities for Low-Income Countries

The Fund´s traditional focus on non-concessional short-term programs to restore macroeconomic balance ended in the mid-1980s[1].  Heavy foreign debt and low economic performance were found to preclude most LICs from borrowing at the usual non-concessional IMF terms. The IMF therefore responded by introducing new facilities designed specifically for LICs. The public quest for more country ownership and for a strengthened focus on poverty reduction led in 1999 to the Poverty Reduction and Growth Facility (PRGF). To increase and improve financial support to low-income countries, creditors also resolved to make greater efforts to restore debt sustainability and fiscal space in LICs by providing comprehensive debt relief, first under the Heavily Indebted Poor Countries (HIPC) Initiative[2]. Multilateral debt could for the first time be written down, including debts owed to the IMF itself. In 2005, the IMF approved the Multilateral Debt Relief Initiative (MDRI) to provide significant further debt reduction beyond the HIPC Initiative.  
The 2009 reforms created a new architecture of concessional facilities aimed at providing more flexible and tailored support to meet the diverse needs of LICs. Since then, the Extended Credit Facility (ECF) the main concessional tool for addressing protracted balance of payments problems. The ECF is the most important of three concessional lending windows that are part of the Poverty Reduction and Growth Trust (PRGT). The Standby Credit Facility (SCF) was created to provide support to LICs with short-term balance of payments needs, with the possibility of using it on a precautionary basis. The Rapid Credit Facility (RCF) was created to provide rapid low-access financing with limited conditionality. Interest rates were reduced to zero on all concessional facilities as a temporary measure in response to the crisis—a measure that has been extended through end-2014. 


Figure 1:  Projected Number of Countries Eligible for PRGT


The number of LICs eligible for support from concessional IMF lending is projected to drop from currently slightly above 70 to 50 by 2050 (Figure 5.1.) Future demand for IMF concessional lending is projected by the Fund to remain elevated, despite real GDP growth for the median LIC projected to grow at around 6 percent. However, strong expected growth performance notwithstanding, vulnerabilities remain elevated in LICs, suggesting the potential need for continued strong IMF support over the medium term. Over the past two years, many LICs have reduced fiscal space and larger current account deficits than prior to the crisis. Increasing financial and trade openness and higher GDP will translate a given shock into higher concessional financing needs until graduation, according to the IMF (OP#277, p. 32). Moreover, particularly fragile states and those emerging from conflict, may still require repeated RCF and ECF support before they are able to strengthen their macroeconomic positions sufficiently that they no longer require longer-term Fund financing (p.33). For the period 2013-35, the IMF (OP#277, Figure 4.2.) projects the average annual demand for PRGT lending at between SDR 1.3bn and 1.7bn ($2.01bn, resp. $2.63bn).


What are the specific strategic options for the next decade? For the Fund´s concessional engagement with low income countries, three strategic options seem particularly worthy of further debate:

  •  Increasing the scope for blending subsidized resources from the PRGT with general resources of the Fund in order to preserve concessional resources for the poorer LICs.  Growing divergence of economic performance and political stability among low income countries should be reflected in a higher share of blend finance going forward[3]. Another important argument for raising the amount of blend finance to LICs is the budget constraints of the traditional donor countries: blending means more leverage per given resources[4]. Currently, PRGT-eligible members are presumed to blend if (i) their per capita GNI exceeds 100 percent of the IDA operational cutoff; or if (ii) their per capita GNI exceeds 80 percent of the IDA operational cutoff and they also have market access as defined below[5]; and (iii) they are not at high risk of debt distress or in debt distress. Increasing the share of blending could be defined as a function of sustained past and prospective future access to non-concessional lending from capital markets and official lenders, including emerging partners; median (not mean) or median/mean per capita GNI, with various percentage points relative to the IDA cutoff, below or above.
  • It is recommended[6] to bring the average grant element in IMF concessional lending to PRGT-only countries closer to those of the soft windows of the MDBs. Until recently, contracting of non-concessional debt, i.e., debt with a grant element of less than 35 percent was prohibited in IMF-supported programs in LICs; while contracting of concessional debt, i.e. debt with a grant element of at least 35 percent, was not subject to any limit. Mainly due to shorter maturities (10 v 40 years), IMF lending has a lower grant element (only now at ca. 35%)[7] than IDA lending (61.1%). The Fund cannot provide outright grants as these would undermine its revolving nature. Raising the IMF grant element would be effected by longer grace and maturity periods in IMF lending; it would also require more resources as the IMF lending capacity would be cut, but this capacity effect could be compensated by a lower number of PRGT-only eligible countries. Should scenarios implying a lower number of PRGT-only countries not materialize, more resources would be needed, however[8].
  • Disaster-poverty hotspots - as identified by ODI (2013)[9] – would particularly benefit from adding an insurance-type instrument with ex ante qualification criteria to the PRGT lending facilities.  The grace period for PRGT lending, currently a fixed period of five years, could be complemented by a five‐year floating grace period, which the country could use whenever it was hit by a predetermined natural shock to exports, GDP, or programmed aid. When LICs can chose between counter cyclical credit lines or disaster-insurance, they will opt for the latter the larger the possible loss (in terms of GDP, say), the lower the cost of insurance, and the higher interest rate the country pays when it borrows. To be sure, current PRGT-refinancing cost can only rise (not fall any longer) up to 2025 as they are close to the zero bound. Projecting rising interest rates, a facility with no initial grace period but with a five‐year floating grace period, which the country could use whenever it was hit by a shock, would become increasingly attractive[10]. 

So far, judging from the publications that the Fund puts in a very transparent manner on its website, the strategic considerations with respect to its future as a source of concessional finance leave the impression that their coordination with other multilateral soft-finance windows ncould be improved upon. This might hold especially when the Fund´s grant element for PRGT-only countries is raised by lengthened lending maturities, bringing its profile closer to those of the multilateral development banks. Different ministerial tutelage, however, will be a stumbling block to closer coordination.

[1] For more historical detail on the IMF history of concessional finance, see Yasemin Bal Gündüz et al. (2013), The Economic Impact of IMF-Supported Programs in Low-Income Countries, IMF Occasional Papers #277.
[2] HIPC trust funds are separate from PRGT funds and will not be discussed here.
[3] Find a detailed discussion of blending options in IMF (2013), Review Of Facilities For Low-Income Countries— Proposals For Implementation, March 2013
[4] The use of windfall profits from gold sales was an important source for subsidizing the PRGT in the recent past. The IMF Executive Board took a decision in September 2012 that would increase annual lending capacity to LICs to an estimated SDR 1.25 billion on a self-sustained basis (i.e., without a further infusion of resources to the PRGT). Absent a persistent change in trends, this is sufficient to meet the average projected base level demand for PRGT resources over the next 20 years, according to OP#277. Note, however, that an earlier IMF review had been less optimistic: “Updated longer-term demand projections suggest that LICs’ financing needs will outstrip the PRGT’s lending capacity.” IMF (2012), Review of Facilities for Low-Income Countries, June, p.18. That review produced a wider margin of PRGT-demand for the same period than the recent OP#277, from SDR 1.2bn to 1.9bn, which may indicate some internal vetting.
[5] The 80% rule currently applies to India and Pakistan (interview with C. Mumssen, 6th March, 2014).
[6] See also Matthew Martin and Alison Johnson (2004), The Role of Multilateral Grants in Financing the MDGs, Commission for Africa Background Paper, Development Finance International: London, September.
[7] IMF, Debt limits in Fund-Supported Programs: Proposed New Guidelines, August 2009 calculated an average grant element of 27.6%, which has been raised to 35% (C. Mumssen, 6th March, 2014.
[8] The formula to compute the budget implications is given by: Grant Element (GE), % = 100(FV – NPV)/FV => NPVIMF/IDA = GEIMF/IDA x NPVIMF =NPVIDA .
[9] ODI (2013), “The geography of climate change, disasters and climate extremes in 2030”, Overseas Development Institute: London.
[10] This facility would follow the example set by the French Agence Française de Développement initially suggested by D. Cohen, P. Jacquet and H. Reisen (2006), After Gleneagles: What Role for Loans in ODA?, OECD Development Centre Policy Brief No. 31, OECD: Paris, December. For how to apply the AFD loans to the IMF concessional facilities, see Eduardo Borensztein et al (2008), “Aid Volatility and Macro Risks in Low-Income Countries”, OECD Development Centre Working Paper No. 273, OECD: Paris, June.

Monday, 3 March 2014

Will ´Smart Aid` Ignore Half of the World´s Poor?

Inequality matters enormously for projected poverty outcomes, as has been demonstrated in a recent study by Edward and Sumner (2013)[1]. It would seem that inequality trends will not matter for ´Smart Aid`, though, if (and only if) OECD donors define aid as “smart aid” increasingly targeted towards the poorest countries (Lomøy, 2014)[2]: “If it’s to succeed, it must become ‘smart’ – increasingly targeted towards the poorest countries and those that face the greatest difficulties in raising alternative finance for development”. If the view expressed by the DCD director reflects DAC views, then the middle-income country classification in itself is a reason for reducing or even ending aid. This would exclude aid to the poorest who (currently) are overwhelmingly in countries ( such as India, China or Indonesia) classified as middle income countries (MICs). The UK Parliament has just released a similar recommendation: “We recommend that, in general, Middle Income Countries should graduate from aid, but in a controlled manner that takes account of needs and resources” (para 20)[3].

But as Edward and Sumner (2013) show this view will neglect the majority of the world´s poor, except under quite optimistic assumptions about both growth and inequality trends (Table 1). Under a scenario – quite realistic as we write (early 2014) – that countries achieve only half the IMF growth forecast (and that is still 3% in real terms) and that current inequality trends continue, 52% of the world´s extreme poor will still live in countries classified as MICs in 2030.

Distribution of $1.25 Poverty, % of World Poverty, 2030

Inequality trend
Current inequality trends
Static inequality
Lowest inequality
Stable MICs
Fragile LICs
Fragile MICs

Source: Edward and Sumner (2013)

OECD donors could only ignore more than half of the world´s extreme poor, if they assumed that the MICs could mobilize enough own resources  to reduce the inequality within their respective countries.  The poor would come under new responsibility, from OECD donors to their own governments. Such an attitude would reflect excessive optimism around about poor countries´ capacity to mobilize their own resources rather than to depend on aid dollars, via redistribution from the “rich” to the “poor” within developing countries.

[1] Peter Edward and Andy Sumner (2013), “The Future of Global Poverty in a Multi-Speed World: New Estimates of Scale and Location, 2010–2030”, CGD Working Paper #327, June.
[2] Jon Lomøy (2014), “Getting Smart with Aid”, OECD Insights, 27th February.
[3] UK Parliament (2014), The Future of UK Development Cooperation: Phase 1:Development Finance, London: The Stationery Office Limited, 5th February.